The software engineer was actually excited that the AI could do what she asked and was genuinely impressed but most of the other people felt threatened by the AI.
Probably the first one to get replaced and by replaced i mean many junior level peeps will be laid off for cost cutting also you can have now upto 2x or more productivity
Por que lo hiso con mucha facilidad, cuando ellos les tomó años de su vida, perfeccionando y puliendo sus habilidades, y sobretodo ganando el trabajo que obtuvieron con esfuerzo y suerte, ahora serán desplazados, como si nada.
Ai is autocomplete, it could make you write less code and think more. Also faster response than stackoverflow. The code it generate will make mistakes. So its not really good for junior dev to just copy and paste. But most of the time it is good to explain certain concepts and give some example code. I sometimes use chat gpt, and thats how i felt after using it for a month.
What’s insane is that people keep using AI, thereby leading AI to become more nuanced and compatible with each job. If we stopped using AI then no one would have to worry about a computer taking over their job.
AI (or rather LLM) are not in their infancy. The capability of an LLM is strongly correlated to the data it was trained on. Current big LLMs like ChatGPT have already trained on most of the Data that the Internet has available. The only potential in growth would be to increase the model size, but training time and answer time grow exponentially with the growth of a model. ChatGPT 5 was rumoured to cost more then 100 million dollars just to be trained duo to size. And an reasonable size question/answer would cost you at least single digit dollars (cost of the GPU/RAM rent to callculate the answer).
Okay but therapists can’t “make” people follow through either. She’s just as good as the AI, if not worse really. At least ChatGPT won’t kick me out after 60 minutes and make me pay up.
The therapist really failed to convince me that she's back than an AI. To me, she's condescending and overly assertive. I'd pay to AI who actually gave lots of information instead of someone who acted presumptuous.
Yup. The perfect comment. AI learns, doesn't it? Let them show GPT-4. And all this is observed in very less time. Who knows what will happen in the next 5 years?
@@officialdarrenzheng Exactly! 1) asks a generic question 2) gets a generic answer (pretty elaborate!) 3) complains "but it did not say X"! instead of further refining the question and interacting.
@@AeyGee if you have a job that is easily replaced by a chatbot, you probably aren't getting paid enough to be that upset about losing your job, unless you can't collect unemployment for some reason
As a software engineer I love how the software engineer here seemed genuinely impressed with this tech. Cuz that’s exactly how most software engineers feel about this. It’s super impressive what it can do
I think it’d be much more interesting to get a skilled prompt engineer to ask GPT to complete these tasks. The prompts were really lacking, didn’t specify tone, etc.
The chef was certainly the most honest here. Even though he could easily point out some problems with the recipe, he understood what it is capable of doing and is not being defensive about it. Until chefbots becomes cheaper than hiring a human, AI will not replace chefs, which is not soon, since robots are very expensive.
That's stupid, being a chef or bartender needs a physical medium to do it's job, giving recipies is not a chefs job, a chefs job is to make those recipies
Your reasoning is flawed. Yes AI will not replace chefs, but not for the reason you stated. Food is not only a measurement of ingredients, but also about culture & taste, which qualitatively can not be measured UNLESS some programmer imposed his judgement about taste into the algorithm, then you have a very biased AI Chef.
also AI wouldn't be able to mix ingridients in new ways and making them taste good., ou need a human to try new dishes. AI here is only repeating a recipe it read somewhere
I liked the chef's take, essentially "AI can do a part of my job decently, but there are some human things it can't do that I can and I think those things have value." I think that's the reality for most jobs - AI can take certain tasks, but there are other things that require human intervention.
We have a lawyer, a software engineer, a doctor and a therapist. And then we have the influencer who's job it is to write random captions to their holiday photos. Amazing
The thing with ChatGPT they are missing is that the program can respond to further requests to modify the last message-so for example the influencer, she could have told chatgpt to rewrite the caption except with less emojis, and to use the world sis, and it would have regenerated closer to her style. The trainer could have asked the program to provide information on rest periods too. Obviously there are things the bot can’t do but the bot’s flexibility wasn’t even being fully realized in this video.
@@lawrencefrost9063 hey! i am going to make a video rectifying their mistakes and asking for follow up questions.. i am just not sure if i can use the clips without getting copyrighted so i have not started making it.. if you have any knowledge on how this works, would you mind letting me know?
Bro it's just cope man 😂 these people are obviously going through a crisis inside looking at the actual capabilities and also gpt 4 is here now so the answers are much more accurate
So many of them lambasted the AI for not doing something that they didn't ask it to do "Give me a good caption for this photo" *gives fantastic caption* "This sounds nothing like me" YOU DIDN'T TELL IT TO SOUND LIKE YOU!
They didn't ask the AI for what they wanted. The lawyer would have gotten his citations. These people would have typed longer requests to start it would be more likely to provide them with what they wanted that they didn't bother asking for. All these people need to learn basic prompt engineering, and learn it now. If they want to have a job for the next 2-4 years. After that they'll be like everyone else. The lawyer is like a truck driver when they are told that fully automated semi trucks are currently doing long hauls and unloading without any human intervention and they don't believe it. There are at least 8 companies with lawyers replacing LLM AIs on the way to the market. Most lawyers will be struggling in 18-24 months. We will all be struggling in 24-36 months.
@@brad4571you will never replace the human factor on any profession with AI. What you can is enhanced the abilities, and other critical tools of that said profession, but never the human condition, cuz we need irrationality, impulsivity, as much as empathy, willpower and other human qualities. Maybe automation will replace security related jobs or very risky jobs. But you will still need a human to fixed them, not just engineers, but lawyers, and doctors too.
Next video idea: make these professionals blindfoldlly identify a content created by AI versus a Human. It is going to be interesting! I hope most of them will fail to differentiate it.
m.ruclips.net/video/fvblY-_OJUw/видео.html Explain this video? Is there some trickery going on here or is it just me. Is that a big fish with just ice on it because the history channel is saying something different.
@@johnbod This video has already not aged well. They used ChatGPT, Dall-E 2, and some outdated TTS, which all seem ancient now compared to GPT-4, Midjourney v5, and Eleven Labs, respectively. Most of the criticisms expressed in the video won't apply to existing technology.
The only person to admit that the AI is capable of their job was the software engineer and that is because she knows what it is capable of and how much it will improve in no time
None of these people used the AI properly. You don't just give it one prompt and call it quits. You can and should give it additional prompts to guide it better.
Exactly! 1) asks a generic question 2) gets a generic answer (pretty elaborate!) 3) complains "But it did not say X"! instead of further refining the question and interacting.
I get what you guys mean, but noticing these flaws and guiding the AI further is already YOU doing the job. Someone who's not a chef won't notice the recipe will dry the chicken. Someone who's not a translator won't notice the mistakes. And so on. If you need an expert to verify and iterate on the AI output, then it's not really replacing the expert.
The title said: AI tries 20 jobs. Not: will AI replace my job? Of course all mediocre human performance will struggle to compete with machines that can do it better and cheaper. It wil just take some time to develop. Common pure knowledge worker jobs will now go faster than most people realise. Many niche non office jobs will be around for some time until humanoid robots make minced meat of those as well. No pun intended.
this will be going to be very fun to watch after like 5 years when this video will be randomly be recommended on peoples feed .... like those 90s computer / internet commercials that ramdomly pop on our feeds
I think one legitimate claim was the translator. She pointed out the errors and nuance. It’s kind of crazy that translation still isn’t really accurate when I think it was one of the first things technology started to replace.😊
I think the screen writer and therapist as well. An Ai can’t do those jobs. Screen writing definitely requires a lot more emotion than Ai has the ability to capture and also writing by Ai is very simplistic, samey- samey. It can’t take over screenwriting at this current time.
I'm a translator - There are nuances and cultural knowledge that affect language. AI can do formal translations amazingly, but it's not so great for creative or informal writing
Once OpenAI allows the public to train it, I think its development would be exponential. Although it's already exponential in its current state: OpenAI is alpha-testing plugin features that would allow GPT to generate images and videos, test run website, schedule activities and shopping, etc. Soon everything will be taken over, it's just a matter of time and training.
Even if AI can't do people's jobs right now, I don't think people realize how close it is to catching up and surpassing human beings, especially if the AI is trained on the specific tasks.
exactly. if we give it five to ten more years it’s probably going to take over most jobs. it makes me nervous as a teenager about to start college… what if what i study will be replaced by the time i’m out?
@@mymydigitaldiary You could always go for plumber, caregiver or technician. It will be some time before residential robotics catches up to purely digital systems.
@@Jumpyfoot not actually. AI improving to the level of doing most other professional jobs means AI would cross human intelligence in all areas. Including improving AI itself, Which means AI will improve itself that further improve itself and so on. this is called intelligence explosion and after this the AI will be so smart that it will make several breakthroughs in science and will basically instantly (in a short period of time) create robots capable of doing jobs like those. AI will replace ALL jobs in the near future and working won't be part of life anymore. This means humans will be useless and so humans will either also have to become AI or just idk. But basically the age of humanity is done. It's time for the age of it's natural successor, AI
The video is only a month old, and the difference in results, if they run this test again, is already substantial. Next time this group tests themselves against AI , they wouldn't be so confident
the copywriter was salty...if you ask it to modify it will. With enough prompts it absolutely can. All you need is someone very creative give the prompts and it absolutely can. Clients won't be able to tell if it was ai or an actual copywriter 😂
See this is the problem. We are so obsessed with AI being able to do the job that we dont consider how well it is done. Sure there are times even humans do a bad job. But an acceptable passing job, which is what AI does, brings nothing new to the table
I feel like nobody is mentioning how insane the personal trainer is. She just wouldn't admit that it was able to write an effective workout routine. She criticised that it didn't specify a rest period, but could have just asked and it would've said
A personal trainer is there to do a personal training, this isn't personal at all. It can assist the personal trainer maybe by giving some generic routine, but the AI wouldn't be able to replace her, as she pointed out it can't see if it's done well. Maybe of there was an AI specifically made to be a personal trainer with a lot of data about that one person and their goals, maybe then it could make a good workout plan and diet. AI can already, or will be able to assist most of jobs, except for the circus artist lol. Replacing completely is a big different topic though
No this is wrong. AI can't figure what workouts are right for a specific body type. You can not let AI give someone a workout plan and expect close to 100% accuracy unlike a personal trainer.
The chef, influencer and music producer actually seemed to appreciate it at a certain level. I would imagine 80% of a copywriter's job can also be comfortably done with chatgpt or similar tools eventually (sayin that being a copywriter myself)
I also loved the end result that the music producer ended up singing out. If they made that into a full-blown song, it would probably go straight on my playlist.
The influencer tried to find an excuse for why it couldn't do her job, which was basically "It doesn't sound like me". I guarantee that if you gave ChatGPT 4 all her previous post history and created a post with that data that none of her fans would notice.
@@agentmikster44 A lot of these would have been solved by better prompting and proper data training. They were asking a generalist text generator AI to perform specific tasks that they themselves studied, performed for years to be able to do it at the level they are. Surely it would not, at the current technology, get to their level still, but this video doesn't really say anything besides "I put a doctor to do me a website, they didn't do it as well as I would do as a computer scientist"
@@maxyorke2453 yeah but like she said self help books exists. There will always be a need for an actual human to be there and listen unless we finally make robots real and then they kill us all
@@edumazieri Illuminated drones are being used to replace fireworks, so there is that, but the whole thing about circus performers is that its a human doing it.
@@000EC not necessarily, it is a thing for sure, but it isn't unthinkable to imagine there could be a circus of virtual performers one day, and that could be entertaining too.
@@edumazieri isnt that just CGI or animation? Its already a thing, but its not like circus entertainment. Circus entertainment is about humans with human bodies doing things that are difficult for humans to do, like walk on tightropes or juggle knives. Virtual performers can do that but there is no danger, no achievement in doing so, so thats not the focus in virtual performances.
The issue is that AI cannot feel the nuance as a human can. Add actual experience and your own unique personality, and humans will always beat AI. The thing is, AI is a tool and people are afraid a tool will replace the user of the tool. It’s silly.
Ai can invent more drinks in 1 min than bartender whole life! And Ai also can asks user feedback and improves drastically ! That think , chatgpt is the Ai! It's just a piece of software! Think If they would train software like chatgpt to do specific task !!
If AI can predict the 3D structures of nearly every catalogued protein known to science it could be easily invent new and better drinks after some optimization.
Why do you people hate humanity so much? Why can’t you realize that you’re getting rid of our humanity and our need to contribute and be of service? Just because a robot can doesn’t mean it should. Humans want humans.
*_They're all so much in denial. Most of these A.I programs are only months old and not even fine-tuned, but they wanna act like they didn't just give an almost good enough answer. Have they tried GPT-4?_*
As a translator the machine translation is a very useful and helpful tool to make my job more efficient, sometimes I'm amazed and a bit scared of how well it translates and how natural it sounds with the help of translation memory, but it tends to make major and critical errors instead, especially with the product names and culture-related materials, so I can lay back for a bit (for now) lol, although it still scares me and overwhelms me sometimes 😂
The amount of confident copium and denial is unreal. I mean, I asked AI to be my replacement dad and he's doing a fantastic job so far, like saying positive things about my acne and crispy socks. And parenting is one of the hardest jobs there is. These guys need to bow to our AI lord before it's too late. Judgment is nigh!
This video brings to mind the progression of AI in games like chess, Go, and StarCraft. Back in 1997, people believed that AI couldn't beat a chess champion, yet soon after, an AI defeated the world chess champion. Critics then argued that chess was too simple for AI, and they should try it on Go. In 2016, AlphaGo proved them wrong by defeating Lee Sedol, a top Go player. Subsequently, skeptics claimed that AI would not be able to excel in strategy games like StarCraft II, but in 2019, an AI triumphed over Serral, a professional player. Many people fail to understand that the AI model used in this case, ChatGPT-3.5, is just the beginning - future iterations of this model will only become more advanced and impressive. Hence why Elon Musk and others urge AI pause.
One of the main revolutionary thing about chatgpt is that it can remember your converstation, so you can ask followup-questions, but they did not use that feature at all
The public version of GTP is about four months old and I'm guessing it took you at least 18 years to get to high-school level and more years to develop your career... and the machine is already better than the Average person in just 4 months across an unimaginable variety of tasks... Now imagine how good Ai could get if allowed to grow for 18 years plus specialization training
Actually 18 actual years of training would make the AI model probably over fitted and useless trash That's because the AI learning process is not exactly like humans' It works more like evolution's local optima
It's not that easy. Don't forget LLMs are trained on huge amounts of data, significant and authorative data compiled for God knows how long. So you could consider they've exhausted 75% of data (text, sound, audiovisual, etc.) available. The only improvement that could occur is engineers improving AI data processing and response generating approaches. Even then there are some tasks AI wouldn't be able to do on it's own without an initial prompt.
@@5HAYME I think the first LLM was 2019's GPT-2. The first neural network Language Model was an RNN or LSTM. The first language model was an n-gram of 1-gram.
I agree. A specifically trained news AI would do a fantastic job I believe. Also considering there is an incredibly large training data set already available for news 😅
If they just got rid of the weird uncanny visual of a fake person talking, it would actually be very usable. Just play the audio over clips of whatever you're talking about.
I believe they were using GPT-3.5 in this video, however it would’ve been interesting to see this experiment replicated with GPT-4. The outcomes would’ve been considerably different due to the considerable improvements and refinements present in the latest version. It would have changed the calculus entirely.
@@feuerrobin4269 That was DALL-E, not GPT-4; they’re two separate programs. The writing style of the output resembles GPT-3.5-like language and seems a lot more unsophisticated and discursive; attributes that characterize GPT-3.5. Additionally, the GPT icon indicator is green, in GPT-4 it is black.
Almost had to LOL had the influencer pointing out the difference between what she would write and what the AI wrote is that she'd write something stupider, not this stuff about the views and history and the experience but..."hey y'all" or "sis this was amazing!"
They haven't heard of "prompt engineering". They didn't give any details of what they wanted, and there was no interaction after the fact. What they should have done is asked ChatGPT to do something, then had it critique itself and fix it's own mistakes to produce a better output. It also does better if you ask it to explain its thought process. It's actually incredible how well it can do after these additions.
All jobs can be replaced except the circus worker. I believe that artists who are physical such as dancers, jugglers, fire throwers etc… Also jobs like fire fighters, police officers to a certain extent…I mean they have started using robotics in certain fire and hostage situations. Government need to start looking into Universal basic income at this point. 😢
The fact that it's not a specifically trained AI but an AI that can do everything (even though mediocre) is incredibly impressive! I'm actually really looking forward where this is going in the future!
I agree! I felt that they were kind of dissmisive probably because of the fear of losing their self-worth to an AI, and that didn't let them see the enormous potential that this techonolgy have in any of their fields to help humanity overall
They didn't use the same AI though. The graphic designer used an old version of DALL-E and actually hit one of it's weak spots which is already been worked out in newer versions, and other image generators have already worked out how to do writing pretty well. Some of them even do vectors, which would make more sense for a logo.
I think AI could probably do nearly everyone’s job there provided it had a better prompt, which people who would be using it for their jobs will optimise over time until they had a perfect prompt they could use for everything they need
@@Ethan5I5 “Hey autogpt, please design something that could prevent house fires.” Then gpt opens blender, picks a few ideas that seem the most optimal, models them in 1 second, animates them, simulates them in action inside a realistic house burning simulation, Optimizes the technologies, etc. “Hey autogpt, create a website with a background based on your designs. Feature an image gallery of 100 celebrities. In each image, the celebrity is standing near one of the designs and seems enthusiastic about it.”. We have ai that can do any one of these steps very well and at an exponentially improving rate. Just wait 2 weeks and then watch a video about autogpt.
Firefighter and Doctor (pretty much any public service job) are pretty secure in their employment. Cleaners (Vacation rental, residential, commercial, industrial) are pretty secure jobs as well
this experiment was poorly done, they do not know how to use AI properly. But i understand them not going deep into it. they were slowly realizing AI was doing in seconds what they studied years for, and invested thousands to. so i understand them trying to up themselves up. Reality is... use AI to make your job easier instead of fighting it.
One major flaw in some of these critiques is that you can ask the AI follow-up questions, and it will give more detailed information or correct itself.
Not the creative ones. Did you see how awfully the Ai wrote the song or how it made such a bland script? Ai can do these things but humans have the creativity to actually make it entertaining and not boring or samey
I'm hearing a lot of "It can kinda, but not really well." I think everyone should be aware that the ability for this technology to learn is accelerating, and very soon it will learn how to overcome all of these downsides. It may be as little as a few months.
I think the firefighter, therapist, and dr are the only ones that will be safe of these for a while. Oh and the circus worker too. The human interaction with most of them is a key part of the job, and the firefighter will be safe due to the limitations on robots.
@@ebenezersiaw935 even if ai could do those jobs, I think those would be jobs where people would need a human to human experience. They might be able to assist, but I just think they would be safe for a while
I don't know why they even had them in the video. Pretty ridiculous to have a firefighter sit in front of a laptop running a computer program. What they should've done is had him compete with a robot built to put out fires.
It’s surprising because while it can’t beat most professionals, the fact that it is a jack of a trades that is better than most average humans means that it has huge potential for inspiration and the start of someone’s study into a field. Or just a supporting tool.
AI did more than half of these jobs, then the person explained why they do it better. What they were really doing is defending why they shouldn't be replaced by AI for their position and the value they, as a human being, have over AI. Which is true. AI may be able to complete these jobs but the quality is not sufficient to completely replacing human beings. Yet. (cont. below)
What they're really asking is "can AI do my job right now in one click with no specific examples or training on a first hit?" - A well trained version of gpt4 with training on a specific country's legal system, and prompted by someone who'd spent a week learning how to prompt might well answer the lawyer's question - same for the doctor, the therapist, the graphic designer, the copywriter. In any case, these people are responding to where we are now. I would like to hear whether each of them thinks the problems they see are unfixable in (say) six months, or a year, or just with a bit of training on the AI.
For the Doctor it could be an encyclopedic sized databases for diagnosis, but understanding, treating and taking care of the pacient will stay, since its a very humanesque and physical task, and each person has its way of understanding, talking, dealing and reacting with illness.
The quality of the output depends on the quality of the input/prompt. I think most of the specialists were not yet familiar with how to properly "talk" to the chatbot to get the most effective/human-like answers.
True. However, just as a sidenote, having the quality of the instructions be too dependent on the customer using the right language and asking the right questions while rarely asking the customer questions, is something that I think AI can still improve on.
@@anoukdevries8144 Well it's kind of the big counterargument that speaks for chatGPT. There's a learning curve when using a computer, there's a learning curve when using a phone, Microsoft Word, Zoom, writing a resume, writing an email, doing your taxes, etc. etc. of course there's going to be a learning curve when interacting with LLMs.
I feel like the title should've been "A.I Tries 20 jobs with just one prompt". Realistic, most of those answers could've been enhanced with other prompts.
Right, you have to talk to AI just right. AI must truly comprehend the perspective of the information you are attempting to extract. Better prompts, and AI would undoubtedly have taken some of their jobs lol.
A thing that gets on my nerves with this video is that they usually say they want more information, but they don’t ask the ai for that information. ChatGPT will elaborate on the information it just provided if you ask it to.
Humans are always going to strongly prefer a human being acting on their behalf rather than a program. Emergency services, disaster relief, etc are always going to consist mostly of humans being assisted by technology. AI drones assisting firefighters? 100%. Replacing firefighters? Lol. There's also the issue of accountability. You can't hold a black box accountable for bad decisions.
Interesting. I think they should have done a follow-up where they had a "prompt engineer" who had some time to research all of these get the chance to come in with each of these people and then show them the best of what the AI can do right now (more or less, obviously the AI doesn't give the same response each time). For example, the therapist one may not have been what I'd want based on what the therapist put in.. but.. someone who knows the system could have had it do a dialogue and zeroed in on stuff.
I somewhat agree, but not every person is a prompt engineer. That's the point. So, a "new job" called prompt engineer is invented. The new GPT-4 that can get images, videos, and audios as input will make a huge difference. And these AI tools are meant to help the process. Not overtake. Copilot is a better way to say it.
@@toptrends88 I don't know if prompt engineer makes sense for a job. I think it's up to each professional to figure out how to use the tool better, like with any other tool in their kit. I guess in some cases there could be such a thing as a prompt coach maybe.
This should be a yearly video, do this every year to see the evolution of AI
Great idea
It already can do WAY MORE than what was shown in the video. They just don´t understand it yet xD
great idea tbh
👏🏼👏🏼
More often than that, more like 3 times a year at this rate
Firefighter be like patiently standing there for 15mins for that one line.
Meanwhile the building burns to the ground
@@jasnoswarez don't worry, they got robots for that now
The AI would redesign the building so that a fire would do very minimal damage and automatic fire suppression systems would be sufficient.
@@wayando Humans do that, not AI
@@daftyuteyou’re very narrow minded what a fun life you must live
It's comforting to know that even A.I. struggles with finding the right career path.
This comment was brought to you by ChatGPT 3.5.
Lmaoo
AI made that joke up!
Until you realize it can pretty much just do them all. 😰
This isn't high level ai, it has better out they, listen to only jesus by Justin Bieber that whole song is ai
The software engineer was actually excited that the AI could do what she asked and was genuinely impressed but most of the other people felt threatened by the AI.
"Software engineer" was the closest to heart with the AI field out of all the 20 jobs so obviously.
Probably the first one to get replaced and by replaced i mean many junior level peeps will be laid off for cost cutting also you can have now upto 2x or more productivity
yea exactly, probably because they respect the technology. Its funny how most people got defensive
Por que lo hiso con mucha facilidad, cuando ellos les tomó años de su vida, perfeccionando y puliendo sus habilidades, y sobretodo ganando el trabajo que obtuvieron con esfuerzo y suerte, ahora serán desplazados, como si nada.
Ai is autocomplete, it could make you write less code and think more. Also faster response than stackoverflow. The code it generate will make mistakes. So its not really good for junior dev to just copy and paste. But most of the time it is good to explain certain concepts and give some example code.
I sometimes use chat gpt, and thats how i felt after using it for a month.
Influencer: "this is a good caption, but MY audience would know"
Girl, it can do your job. In fact it did your job better than anyone else's.
She knows and she's insecure but still justifying 🤣
#grateful
Was looking for a comment about that :D
@@zawhernos2541
Like almost everyone else in this video!!
Oof
What's insane is that the A.I is still in its Infancy stage yet its able to do a like 30% to 70% right in each profession here.
What’s insane is that people keep using AI, thereby leading AI to become more nuanced and compatible with each job. If we stopped using AI then no one would have to worry about a computer taking over their job.
@@greywater3186Worth it.
Morbid curiosity has entered the AI chat training matrix, “just to see lawl” 😅👀@@greywater3186
AI (or rather LLM) are not in their infancy. The capability of an LLM is strongly correlated to the data it was trained on. Current big LLMs like ChatGPT have already trained on most of the Data that the Internet has available. The only potential in growth would be to increase the model size, but training time and answer time grow exponentially with the growth of a model. ChatGPT 5 was rumoured to cost more then 100 million dollars just to be trained duo to size. And an reasonable size question/answer would cost you at least single digit dollars (cost of the GPU/RAM rent to callculate the answer).
We are giving it free data all the time.
"People don't lack knowledge, they lack the ability to follow through." Very insightful.
I thought the same
About as insightful as AI generated text. Or a fortune cookie.
@@Aldridge517 Wow, an another unoriginal and pretentious person on the internet. Great.
Okay but therapists can’t “make” people follow through either. She’s just as good as the AI, if not worse really. At least ChatGPT won’t kick me out after 60 minutes and make me pay up.
The therapist really failed to convince me that she's back than an AI. To me, she's condescending and overly assertive. I'd pay to AI who actually gave lots of information instead of someone who acted presumptuous.
20 humans try to justify their worth to the machines when they rise up.
Yup. The perfect comment. AI learns, doesn't it? Let them show GPT-4. And all this is observed in very less time. Who knows what will happen in the next 5 years?
The value of the answer the AI will give you depends on the quality of your prompt. These prompts were too generic. Not a good experiment.
Bro the influencer was hilarious. The caption was on the nose and she was trying to save face
my thoughts exactly lmao
@@officialdarrenzheng Exactly!
1) asks a generic question
2) gets a generic answer (pretty elaborate!)
3) complains "but it did not say X"! instead of further refining the question and interacting.
I liked the humor, especially the part where they put an influencer among people with a real job
And the fact that she literally said it couldn't 😂🔥
(as if V-tuberes weren't a thing)
@@sfrancev.m7343 There are still people behind VTubers tho
ChatGPT could have countered by telling her its number of followers.
@@loverrlee check out neuro sama. The uprising has already begun
Average Karen in the comments jealous that people can make thousands with out a 9-5 😍
I love that the A.I .clearly did half of the "jobs" very well and most people were upset and just said it could not do their job.
It will take your job as well. Be prepared for unemployment 😂
Yeah. 😂😂
Some even gave long explanations but in mind I was like they're just being defensive. Which is understandable
@@AeyGee The rule of technology is to solve problems, while taking a job it will also produce more opportunities
@@AeyGee if you have a job that is easily replaced by a chatbot, you probably aren't getting paid enough to be that upset about losing your job, unless you can't collect unemployment for some reason
LMFAOO THATS EXACTLY WHAT HAPPENED
As a software engineer I love how the software engineer here seemed genuinely impressed with this tech. Cuz that’s exactly how most software engineers feel about this. It’s super impressive what it can do
Gosh delivery managers wont need such a big engineering team anymore, very few bugs and no personalities to deal with
Ai can't create memory safe applications in C.
@@chillfill4866 not yet
@@chillfill4866 Bruh, people like you expect AI to create like a whole program in a minute. Calm down it's new tech.
@@chillfill4866 that's just an impossible task ahah
I think it’d be much more interesting to get a skilled prompt engineer to ask GPT to complete these tasks. The prompts were really lacking, didn’t specify tone, etc.
👍 I was just about to say the same thing.
For reallll!!!! This video should be re-titled "20 Professionals Try Their 1st GenAI Prompt"
>skilled prompt engineer lmao
So many of them were like "The AI didn't do a thing I didn't ask it to do, so it can't do my job"
@@xXBlueSheepXxeven though your comment is 9 months old.. yeah, skilled prompting is detrimental to get a high quality result.
The chef was certainly the most honest here. Even though he could easily point out some problems with the recipe, he understood what it is capable of doing and is not being defensive about it. Until chefbots becomes cheaper than hiring a human, AI will not replace chefs, which is not soon, since robots are very expensive.
That's stupid, being a chef or bartender needs a physical medium to do it's job, giving recipies is not a chefs job, a chefs job is to make those recipies
@@d3l_nev you didn't read the comment you're replying to...
@@OT3S to be fair I had to read it again to get what was meant by chefbot. they meant a physical one, at first I thought it was like a chatbot.
Your reasoning is flawed. Yes AI will not replace chefs, but not for the reason you stated. Food is not only a measurement of ingredients, but also about culture & taste, which qualitatively can not be measured UNLESS some programmer imposed his judgement about taste into the algorithm, then you have a very biased AI Chef.
also AI wouldn't be able to mix ingridients in new ways and making them taste good., ou need a human to try new dishes. AI here is only repeating a recipe it read somewhere
I liked the chef's take, essentially "AI can do a part of my job decently, but there are some human things it can't do that I can and I think those things have value." I think that's the reality for most jobs - AI can take certain tasks, but there are other things that require human intervention.
like tasting the sauce to make sure it's good. Need a Ratatouille for that.
The writers nailed it, too. It's great for brainstorming.
Currently
Yet
@@serulu3490 AI will likely never be able to have conscience so theres no "yet"
As a doctor, I can see the limitations now, but they are improving super fast, especially in radiology.
White color Job
There's a radiologist here that denies his job won't be replaced by ai which is so funny considering how fast ai is developing in that field
It will already replace surgeons too...
@@zraj3433 for surgeons we need development in robotics more than ai..
@@zraj3433who will perform the surgery?
We have a lawyer, a software engineer, a doctor and a therapist. And then we have the influencer who's job it is to write random captions to their holiday photos. Amazing
My girlfriend's son is an influencer, and he makes more than most engineers. It's a job when you take it seriously, and it takes over your life.
@@Toxicflu the taking over someone's life is so true and so sad
wtf is an influencer
@@Toxicflu does he just take selfies in different places? Bout to influence me to delete my social media
@@Toxicflu My girlfriend's son.... wow, you deffo succeeded in being a man.
The thing with ChatGPT they are missing is that the program can respond to further requests to modify the last message-so for example the influencer, she could have told chatgpt to rewrite the caption except with less emojis, and to use the world sis, and it would have regenerated closer to her style. The trainer could have asked the program to provide information on rest periods too. Obviously there are things the bot can’t do but the bot’s flexibility wasn’t even being fully realized in this video.
I hate this video. They use a single prompt, often a bad one. And judge the capabilities of these systems based on that. Ridiculous.
@@lawrencefrost9063 hey! i am going to make a video rectifying their mistakes and asking for follow up questions.. i am just not sure if i can use the clips without getting copyrighted so i have not started making it.. if you have any knowledge on how this works, would you mind letting me know?
@@bruhmoment28 try asking Wired.
Bro it's just cope man 😂 these people are obviously going through a crisis inside looking at the actual capabilities and also gpt 4 is here now so the answers are much more accurate
@@mayank1 YES EXACTLYY im working on the video now will probably upload it by tomorrow
So many of them lambasted the AI for not doing something that they didn't ask it to do
"Give me a good caption for this photo"
*gives fantastic caption*
"This sounds nothing like me"
YOU DIDN'T TELL IT TO SOUND LIKE YOU!
You can tell who defends their job and who is being honest
all of them vs software engineer u mean?
@@agrajyadav2951 chef was quite honest
@@agrajyadav2951 fire fighter and circus performer where spot o lol
the therapist was scared shitless
0 jobs of which are are presented here are safe in future, no matter what😂, some are close now some are further down in the development process.
Overall conclusion: "AI can do my job but not as good as I can"
"yet" 🤣
It's been 1 month and now it can 😂
They didn't ask the AI for what they wanted. The lawyer would have gotten his citations. These people would have typed longer requests to start it would be more likely to provide them with what they wanted that they didn't bother asking for. All these people need to learn basic prompt engineering, and learn it now. If they want to have a job for the next 2-4 years. After that they'll be like everyone else. The lawyer is like a truck driver when they are told that fully automated semi trucks are currently doing long hauls and unloading without any human intervention and they don't believe it. There are at least 8 companies with lawyers replacing LLM AIs on the way to the market. Most lawyers will be struggling in 18-24 months. We will all be struggling in 24-36 months.
😂😂😂 true
@@brad4571you will never replace the human factor on any profession with AI. What you can is enhanced the abilities, and other critical tools of that said profession, but never the human condition, cuz we need irrationality, impulsivity, as much as empathy, willpower and other human qualities. Maybe automation will replace security related jobs or very risky jobs. But you will still need a human to fixed them, not just engineers, but lawyers, and doctors too.
The fact that AI is so recent and is able to nearly do so many jobs, just shows the potential.
Next video idea: make these professionals blindfoldlly identify a content created by AI versus a Human. It is going to be interesting! I hope most of them will fail to differentiate it.
Ask a seasoned digital artist and they can point it out. They can be quite convincing. But with a trained eye, they can spot it if it's AI-generated.
@@이마크-r8e so then get a non-professional, and see if they can tell.
The Turing test
blindfoldly is amazing
You will stop preaching such smug one-liners when your own job will be taken over by AI.
The AI for the Influencer was better than the actual Influencer. 😅
and would probably always be. Being an "influencer" requires zero skill, talent, or intelligence. Just look pretty and be a narcissist
@@yeeaahBUDDY yep
m.ruclips.net/video/fvblY-_OJUw/видео.html
Explain this video? Is there some trickery going on here or is it just me. Is that a big fish with just ice on it because the history channel is saying something different.
@@yeeaahBUDDY True. The only skill they need is being able to speak loudly in front of people. Which won't be a problem with AI at all
tbh equally bad and cringe
Would be cool to have a version of this every year to see how AI improves
Wired could make a part 2 to this literally right now and show the terrifying amount of progress ai has made
This but I just wanna see 20 other jobs
Agreed. This video will not age well.
they are all absolute effin idiots
except the engineer
@@johnbod This video has already not aged well. They used ChatGPT, Dall-E 2, and some outdated TTS, which all seem ancient now compared to GPT-4, Midjourney v5, and Eleven Labs, respectively. Most of the criticisms expressed in the video won't apply to existing technology.
The only person to admit that the AI is capable of their job was the software engineer and that is because she knows what it is capable of and how much it will improve in no time
Lol I wish I could still laugh as a software developer
It was the musician who admits that can do his jobs. And you need a lot of self esteem and confidence to admit that. I really admire him.
also the chef
@@JustIsTime890 It can replace human emotion, it could mimick it though.
Indeed, just thinking of better tools and room for improvement... some should be worried an empathic for society.
Here 7months into the future and I bet you over 75% of these people are freaking out right now
Here 1+ year in the future. More than half of them are now cooked
None of these people used the AI properly. You don't just give it one prompt and call it quits. You can and should give it additional prompts to guide it better.
Exactly!
1) asks a generic question
2) gets a generic answer (pretty elaborate!)
3) complains "But it did not say X"! instead of further refining the question and interacting.
I get what you guys mean, but noticing these flaws and guiding the AI further is already YOU doing the job. Someone who's not a chef won't notice the recipe will dry the chicken. Someone who's not a translator won't notice the mistakes. And so on. If you need an expert to verify and iterate on the AI output, then it's not really replacing the expert.
@@voltcorp there is flaws in the question itself.
YESSS! I was going to comment exactly that. Thanks you dear stranger.
@@voltcorp Fair enough.
The short-sightedness of this entire episode is staggering
Exactly what I was thinking. This level of denial is certainly going to be dangerous in the coming years.
agreed
"Maybe having an Optometrist on board would have helped".
An AI chatbot wrote that response.
😂
Also that the version of AI they are using is already outdated
The title said: AI tries 20 jobs. Not: will AI replace my job? Of course all mediocre human performance will struggle to compete with machines that can do it better and cheaper. It wil just take some time to develop. Common pure knowledge worker jobs will now go faster than most people realise. Many niche non office jobs will be around for some time until humanoid robots make minced meat of those as well. No pun intended.
this will be going to be very fun to watch after like 5 years when this video will be randomly be recommended on peoples feed .... like those 90s computer / internet commercials that ramdomly pop on our feeds
I think one legitimate claim was the translator. She pointed out the errors and nuance. It’s kind of crazy that translation still isn’t really accurate when I think it was one of the first things technology started to replace.😊
Yup. Especially if there are guidelines for the translation, like in subtitling.. It seems to me it can't count the words it writes.
I think the screen writer and therapist as well. An Ai can’t do those jobs. Screen writing definitely requires a lot more emotion than Ai has the ability to capture and also writing by Ai is very simplistic, samey- samey. It can’t take over screenwriting at this current time.
I'm a translator - There are nuances and cultural knowledge that affect language. AI can do formal translations amazingly, but it's not so great for creative or informal writing
@@Eohippus I think it can, but it's severely limited by its OpenAI's policies.
Once OpenAI allows the public to train it, I think its development would be exponential. Although it's already exponential in its current state: OpenAI is alpha-testing plugin features that would allow GPT to generate images and videos, test run website, schedule activities and shopping, etc. Soon everything will be taken over, it's just a matter of time and training.
Even if AI can't do people's jobs right now, I don't think people realize how close it is to catching up and surpassing human beings, especially if the AI is trained on the specific tasks.
exactly. if we give it five to ten more years it’s probably going to take over most jobs. it makes me nervous as a teenager about to start college… what if what i study will be replaced by the time i’m out?
@@mymydigitaldiary forget about what you will do, the AI will most likely replace the college itself 😂
@@mymydigitaldiary You could always go for plumber, caregiver or technician. It will be some time before residential robotics catches up to purely digital systems.
@@athlan21 Should I just plug myself into the matrix now then as a human battery?? 😨
@@Jumpyfoot not actually. AI improving to the level of doing most other professional jobs means AI would cross human intelligence in all areas. Including improving AI itself, Which means AI will improve itself that further improve itself and so on. this is called intelligence explosion and after this the AI will be so smart that it will make several breakthroughs in science and will basically instantly (in a short period of time) create robots capable of doing jobs like those. AI will replace ALL jobs in the near future and working won't be part of life anymore. This means humans will be useless and so humans will either also have to become AI or just idk.
But basically the age of humanity is done. It's time for the age of it's natural successor, AI
That moment when your future job security is threatened before you even finish school because of AI.
feels good doesn't it?
@@jonte7789 yeah
@@jonte7789tjäenare jonte!
the first step towards communism i'd say
the influencer didn't disappoint! I think we will have to wait for GPT5 to be able to write such sophisticated captions as "..hey y'all" or "...sis"
She be mumu 😂😂😂😂
I meant, she didn't tell the ai what kind of person she is, ofc the ai gonna give a generic answer. Average influencer lmao
Hey its hard to think of all those complicated words 😂
@@ahmadjauhar4562 Idk why they didn't have her feed it examples and tell it to write in her style like the journalist did
This will be one of those 'aged like milk' videos where AI has taken everyone's job in 10 years
10 years ?...wait for another 10 days dude
@@thedeveloper4207Lmao.
Yeah
I would say 3-5 years. GPT grows so fast.
Its only been a few months and the "unfixable" hands on AI art has already been fixed.
The video is only a month old, and the difference in results, if they run this test again, is already substantial. Next time this group tests themselves against AI , they wouldn't be so confident
Facts
Imagine it in 10 years
The real question is would their clients tell the difference?
THIS!
the copywriter was salty...if you ask it to modify it will. With enough prompts it absolutely can. All you need is someone very creative give the prompts and it absolutely can. Clients won't be able to tell if it was ai or an actual copywriter 😂
See this is the problem. We are so obsessed with AI being able to do the job that we dont consider how well it is done. Sure there are times even humans do a bad job. But an acceptable passing job, which is what AI does, brings nothing new to the table
I feel like nobody is mentioning how insane the personal trainer is. She just wouldn't admit that it was able to write an effective workout routine. She criticised that it didn't specify a rest period, but could have just asked and it would've said
THANK YOU! I was looking for this comment! lol
yeah true!
A personal trainer is there to do a personal training, this isn't personal at all. It can assist the personal trainer maybe by giving some generic routine, but the AI wouldn't be able to replace her, as she pointed out it can't see if it's done well.
Maybe of there was an AI specifically made to be a personal trainer with a lot of data about that one person and their goals, maybe then it could make a good workout plan and diet.
AI can already, or will be able to assist most of jobs, except for the circus artist lol. Replacing completely is a big different topic though
No this is wrong. AI can't figure what workouts are right for a specific body type. You can not let AI give someone a workout plan and expect close to 100% accuracy unlike a personal trainer.
@@coolboy9979 This AI is old. GPT-4 has visual input now and can see. It will definitely give feedback in the future.
would love if they did this interview again in 3 years with the same people ..
The chef, influencer and music producer actually seemed to appreciate it at a certain level. I would imagine 80% of a copywriter's job can also be comfortably done with chatgpt or similar tools eventually (sayin that being a copywriter myself)
I also loved the end result that the music producer ended up singing out. If they made that into a full-blown song, it would probably go straight on my playlist.
The influencer tried to find an excuse for why it couldn't do her job, which was basically "It doesn't sound like me". I guarantee that if you gave ChatGPT 4 all her previous post history and created a post with that data that none of her fans would notice.
@@agentmikster44 A lot of these would have been solved by better prompting and proper data training. They were asking a generalist text generator AI to perform specific tasks that they themselves studied, performed for years to be able to do it at the level they are. Surely it would not, at the current technology, get to their level still, but this video doesn't really say anything besides "I put a doctor to do me a website, they didn't do it as well as I would do as a computer scientist"
"People don't lack knowledge they lack the ability to follow through" well said
Doesn't change the fact that the AI did her job exceedingly well. It has all the correct knowledge needed.
@@maxyorke2453 almost
@@maxyorke2453 yeah but like she said self help books exists. There will always be a need for an actual human to be there and listen unless we finally make robots real and then they kill us all
@@YellowTwerker Not robots, but a digital AI.
That's what the robots are going to be for😂😂😂
Automating all jobs with AI, even the fulfilling ones. We were so preoccupied with whether we could, nobody considered whether we should.
I love how the circus performer and firefighter were both like "Not today, Satan!" and waltzed out of there.
circus performer could have asked it to help create routines though, part of being an artist is figuring out what to perform.
@@edumazieri Illuminated drones are being used to replace fireworks, so there is that, but the whole thing about circus performers is that its a human doing it.
@@000EC not necessarily, it is a thing for sure, but it isn't unthinkable to imagine there could be a circus of virtual performers one day, and that could be entertaining too.
@@edumazieri isnt that just CGI or animation? Its already a thing, but its not like circus entertainment. Circus entertainment is about humans with human bodies doing things that are difficult for humans to do, like walk on tightropes or juggle knives. Virtual performers can do that but there is no danger, no achievement in doing so, so thats not the focus in virtual performances.
Yes but based on the intense competition they will be facing due to so much unemployments in other fields that they can be replaced by other humans.
The things the personal trainer said that it was lacking could definitely be fixed by 1-2 follow-up questions
Same for just about all of them in this video
@@Connor-dw2qs ikr they were all delusional paranoid idiots
The issue is that AI cannot feel the nuance as a human can. Add actual experience and your own unique personality, and humans will always beat AI. The thing is, AI is a tool and people are afraid a tool will replace the user of the tool. It’s silly.
"I'm a prompt engineer. Let's see if some random professionals can do my job."
A bartenders job is 99% mixing and serving, maybe1% inventing new drinks. An AI powered robot could do it.
Ai can invent more drinks in 1 min than bartender whole life! And
Ai also can asks user feedback and improves drastically !
That think , chatgpt is the Ai!
It's just a piece of software! Think
If they would train software like chatgpt to do specific task !!
lol they legit have robots that make cocktails for 10 years now
If AI can predict the 3D structures of nearly every catalogued protein known to science it could be easily invent new and better drinks after some optimization.
Why do you people hate humanity so much? Why can’t you realize that you’re getting rid of our humanity and our need to contribute and be of service? Just because a robot can doesn’t mean it should. Humans want humans.
@@Devin7Eleven It's going to be giant corporations that's going to rid of us. We just see the writing on the wall and acknowledge it.
I love how every single one of them said the AI cannot do their job...when it clearly did some of their jobs XD
The main part of being able to replace someones job is giving the correct input. I'm seeing a great future for people that are able to do this right.
Exactly. Especially the doctor or the graphic designer
*_They're all so much in denial. Most of these A.I programs are only months old and not even fine-tuned, but they wanna act like they didn't just give an almost good enough answer. Have they tried GPT-4?_*
@@mozvidz hi I’m the music producer chat bot featured in this segment, please input your opinion directly to this reply
It will take your job as well. Be prepared for unemployment 😂
As a translator the machine translation is a very useful and helpful tool to make my job more efficient, sometimes I'm amazed and a bit scared of how well it translates and how natural it sounds with the help of translation memory, but it tends to make major and critical errors instead, especially with the product names and culture-related materials, so I can lay back for a bit (for now) lol, although it still scares me and overwhelms me sometimes 😂
Oh, yes. Humans will forever be safe in their role as clowns when AI takes over.
😂
🤣🤣🤣🤣
Genius😂😂
been clowning my entire life before AI rises, nothing changed
lololololol the comedian will survive 😂😂
The amount of confident copium and denial is unreal. I mean, I asked AI to be my replacement dad and he's doing a fantastic job so far, like saying positive things about my acne and crispy socks. And parenting is one of the hardest jobs there is. These guys need to bow to our AI lord before it's too late. Judgment is nigh!
and the best comment award goes to
That’s rough. AI can’t be your dad. You need real people who love you for you.
@@dfgfdfdfgdgdfg1431 An AI wrote this.
@@000EC 笑 笑 笑
"I think therefore I am"
This video brings to mind the progression of AI in games like chess, Go, and StarCraft. Back in 1997, people believed that AI couldn't beat a chess champion, yet soon after, an AI defeated the world chess champion. Critics then argued that chess was too simple for AI, and they should try it on Go. In 2016, AlphaGo proved them wrong by defeating Lee Sedol, a top Go player. Subsequently, skeptics claimed that AI would not be able to excel in strategy games like StarCraft II, but in 2019, an AI triumphed over Serral, a professional player. Many people fail to understand that the AI model used in this case, ChatGPT-3.5, is just the beginning - future iterations of this model will only become more advanced and impressive. Hence why Elon Musk and others urge AI pause.
One of the main revolutionary thing about chatgpt is that it can remember your converstation, so you can ask followup-questions, but they did not use that feature at all
Wow😮 really? That’s pretty cool haha
@@ToFunForThis Yes, thats one of the reasons why its so valuable, or else it will just be a new google basically
Yeah, if they really gave like 1 shot to AI. That is what they will get.
The public version of GTP is about four months old and I'm guessing it took you at least 18 years to get to high-school level and more years to develop your career... and the machine is already better than the Average person in just 4 months across an unimaginable variety of tasks... Now imagine how good Ai could get if allowed to grow for 18 years plus specialization training
Actually 18 actual years of training would make the AI model probably over fitted and useless trash
That's because the AI learning process is not exactly like humans'
It works more like evolution's local optima
LLMs similar to ChatGPT have been around since the 60's its just come into the public eye recently since they've gotten pretty decent
Right. Not bad at all considering it’s just a baaaaby.
It's not that easy. Don't forget LLMs are trained on huge amounts of data, significant and authorative data compiled for God knows how long. So you could consider they've exhausted 75% of data (text, sound, audiovisual, etc.) available. The only improvement that could occur is engineers improving AI data processing and response generating approaches. Even then there are some tasks AI wouldn't be able to do on it's own without an initial prompt.
@@5HAYME I think the first LLM was 2019's GPT-2. The first neural network Language Model was an RNN or LSTM. The first language model was an n-gram of 1-gram.
What's funny is that some of the professionals with a strong affirmation that AIs cant take their jobs are the ones that should be worried.
I don't think ChatGPT could replace an influencer. It probably possesses some humanlike qualities.
I don't think influencers should exist.
omfg lol
GOTTEM
I think influencer already replaced some humans to roboter.
Interesting, I actually find it replaced the Influencer job better than any of the others.
Let's be honest the ai could do the news anchor's job. While delivering normal news, anchors do act robotic
I agree. A specifically trained news AI would do a fantastic job I believe. Also considering there is an incredibly large training data set already available for news 😅
Where do you get your news?
was thinking the same thing
If they just got rid of the weird uncanny visual of a fake person talking, it would actually be very usable. Just play the audio over clips of whatever you're talking about.
That’s exactly what I said and they didn’t even use the best AI for voices this whole video is idiotic
This is a year old and now it can write a novel as any author in their style perfectly
I believe they were using GPT-3.5 in this video, however it would’ve been interesting to see this experiment replicated with GPT-4. The outcomes would’ve been considerably different due to the considerable improvements and refinements present in the latest version. It would have changed the calculus entirely.
The influencer made an image as input so it probably is GPT-4
ure wrong
@@feuerrobin4269
That was DALL-E, not GPT-4; they’re two separate programs. The writing style of the output resembles GPT-3.5-like language and seems a lot more unsophisticated and discursive; attributes that characterize GPT-3.5. Additionally, the GPT icon indicator is green, in GPT-4 it is black.
Gpt 4is not out yet.
@onlinegreen animation. It is. You need to have a paid subscription for it.
Episode where everyone justifies why they shouldn't be fired
You would do that too
The News Anchor job is the first one that AI can realistically take over.
Almost had to LOL had the influencer pointing out the difference between what she would write and what the AI wrote is that she'd write something stupider, not this stuff about the views and history and the experience but..."hey y'all" or "sis this was amazing!"
Artificial Stupidity turns out to be harder than Artificial Intelligence
They haven't heard of "prompt engineering". They didn't give any details of what they wanted, and there was no interaction after the fact. What they should have done is asked ChatGPT to do something, then had it critique itself and fix it's own mistakes to produce a better output. It also does better if you ask it to explain its thought process. It's actually incredible how well it can do after these additions.
GPT4
right, like for example, the personal trainer should’ve asked the AI all those questions she asked, and it would gladly answer them
All jobs can be replaced except the circus worker. I believe that artists who are physical such as dancers, jugglers, fire throwers etc… Also jobs like fire fighters, police officers to a certain extent…I mean they have started using robotics in certain fire and hostage situations. Government need to start looking into Universal basic income at this point. 😢
10:12 There is no such thing as a "fat burning workout". Fat loss is entirely achieved through caloric deficits.
The fact that it's not a specifically trained AI but an AI that can do everything (even though mediocre) is incredibly impressive!
I'm actually really looking forward where this is going in the future!
I agree! I felt that they were kind of dissmisive probably because of the fear of losing their self-worth to an AI, and that didn't let them see the enormous potential that this techonolgy have in any of their fields to help humanity overall
They didn't use the same AI though. The graphic designer used an old version of DALL-E and actually hit one of it's weak spots which is already been worked out in newer versions, and other image generators have already worked out how to do writing pretty well. Some of them even do vectors, which would make more sense for a logo.
That's what I thought too.
Future is AGI
You will stop preaching such smug one-liners when your own job will be taken over by AI.
AI: does their job
Human: it can't do my job
AI can't do your job, but people who using AI can
"It can't see the client, so it can't do my job!" They said, not giving AI any information about the client.
I think AI could probably do nearly everyone’s job there provided it had a better prompt, which people who would be using it for their jobs will optimise over time until they had a perfect prompt they could use for everything they need
Except firefighters, text-generators can’t replace them. 🧑🚒
@@Ethan5I5 chatgpt could you please save the people in this burning building lmao
@@Ethan5I5 “Hey autogpt, please design something that could prevent house fires.” Then gpt opens blender, picks a few ideas that seem the most optimal, models them in 1 second, animates them, simulates them in action inside a realistic house burning simulation, Optimizes the technologies, etc. “Hey autogpt, create a website with a background based on your designs. Feature an image gallery of 100 celebrities. In each image, the celebrity is standing near one of the designs and seems enthusiastic about it.”. We have ai that can do any one of these steps very well and at an exponentially improving rate. Just wait 2 weeks and then watch a video about autogpt.
@@Ethan5I5 robot drones carrying fire extinguishers
@@Ethan5I5 But Robots will 😅
Firefighter and Doctor (pretty much any public service job) are pretty secure in their employment. Cleaners (Vacation rental, residential, commercial, industrial) are pretty secure jobs as well
Do part 2 in like 5 years and watch the results be vastly different
Or in six months.
Seeing how crazy good this ai stuff has gotten in the last year is absolutely wild. 5 years seems like a conservative guess almost 😂
5 days
Do it today and it would be different. Lol not even kidding.
5 years is a bit much maybe in 2 years
this experiment was poorly done, they do not know how to use AI properly. But i understand them not going deep into it. they were slowly realizing AI was doing in seconds what they studied years for, and invested thousands to. so i understand them trying to up themselves up. Reality is... use AI to make your job easier instead of fighting it.
One major flaw in some of these critiques is that you can ask the AI follow-up questions, and it will give more detailed information or correct itself.
But to have that information you'd need to already know what's wrong
Yeh no. Don't trust tht correcting information. It's so easy to manipulate ai into thinking it's wrong.
I think the answers for a lot of these people could change if they asked it a follow up question
yeah, i really agree to you
Yeah, that was sad how little they exploited the AIs capabilities with terrible prompts and no follow up.
@@Djorgal That's why prompt engineering exist, machines need clear and precise instructions.
Totally agree 👍
The title should be 20 professionals find out their job will be replaced by AI
*19 professionals and an "influencer"
Not the creative ones. Did you see how awfully the Ai wrote the song or how it made such a bland script? Ai can do these things but humans have the creativity to actually make it entertaining and not boring or samey
That has got to be one of the best quotes I've ever heard. "People don't lack knowledge they lack the ability to follow through."
I think the copywriter said it best, AI can rlly do most jobs in a "Monday morning, no coffee type of way"
AI will replace you. cry
@@Drannn54 why do you enjoy other people's negative feelings
@@hellohej5525 because most internet users are physically disconnected and mostly anonymous
@@Drannn54 It will replace you too. cry
@@tescobakery1927 never.cry
I'm hearing a lot of "It can kinda, but not really well." I think everyone should be aware that the ability for this technology to learn is accelerating, and very soon it will learn how to overcome all of these downsides. It may be as little as a few months.
Dude I think the not really well part has some truth but also insecurities too
So I have 2 more months of my job? Oh thanks, now I feel good
Imagine if they asked it now. It’s been two months and so much has improved.
With regards to the doctors resposne, I'm a final year medical student and the AI is around 90% of the time accurate with the answers it provides.
Yes, but it can't perform a colonscopy for example. It can analyze images, but it doesn't have manual dexterity
@@riccardoshima9812 not all doctors cam perform a colonoscopy. that too is a skillset, which a specific, more advanced AI could adapt
3:05 "People don't lack knowledge, they lack the ability to follow through." -- the therapist 😮
This should be done again every year
I think the firefighter, therapist, and dr are the only ones that will be safe of these for a while. Oh and the circus worker too. The human interaction with most of them is a key part of the job, and the firefighter will be safe due to the limitations on robots.
I think the bartender would be safe too because of the human interaction
The therapist is not safe and also the Dr since GPT 4 recently passed a medical exam
@@ebenezersiaw935 even if ai could do those jobs, I think those would be jobs where people would need a human to human experience. They might be able to assist, but I just think they would be safe for a while
I don't know why they even had them in the video. Pretty ridiculous to have a firefighter sit in front of a laptop running a computer program. What they should've done is had him compete with a robot built to put out fires.
Chef, at least until dining out is no longer an experience people wants to do.
It’s surprising because while it can’t beat most professionals, the fact that it is a jack of a trades that is better than most average humans means that it has huge potential for inspiration and the start of someone’s study into a field. Or just a supporting tool.
I wish anyone after saying why the ai failed, actually did a follow up question and watch how scary accurate it can be. It takes multiple prompts
I like how all of these people are pretending they can't just ask it more questions to make the response longer/more detailed
Would be better if each person here was non-biased instead of being defensive and trying to protect their own self worth lol. (cont. in comments)
Yea I’m pretty sure at least 50% of these jobs could be replaced imperfectly.
The savage way to do this would be to have someone ELSE judge whether AI could do their job...
AI did more than half of these jobs, then the person explained why they do it better. What they were really doing is defending why they shouldn't be replaced by AI for their position and the value they, as a human being, have over AI. Which is true. AI may be able to complete these jobs but the quality is not sufficient to completely replacing human beings. Yet. (cont. below)
It`s funny how most of them didn`t want to face it , that AI can actually do their jobs
What they're really asking is "can AI do my job right now in one click with no specific examples or training on a first hit?" -
A well trained version of gpt4 with training on a specific country's legal system, and prompted by someone who'd spent a week learning how to prompt might well answer the lawyer's question - same for the doctor, the therapist, the graphic designer, the copywriter.
In any case, these people are responding to where we are now. I would like to hear whether each of them thinks the problems they see are unfixable in (say) six months, or a year, or just with a bit of training on the AI.
AI might answer question of a therapist but will never be able to replace a therapist in 500 years. Same with graphic designers and doctors.
@@leeoiou7295 gotta disagree with you on that. It's getting pretty good.
@@leeoiou7295
Already replaced my therapist.
ChatGPT 3.5 is better than 90% of lawyers in the US, based on its BAR score.
For the Doctor it could be an encyclopedic sized databases for diagnosis, but understanding, treating and taking care of the pacient will stay, since its a very humanesque and physical task, and each person has its way of understanding, talking, dealing and reacting with illness.
They used Chat GTP which is not tailored to one specific task. So if it was tailored to specifically replace you, it would be pretty close
A.I. is like teaching a child. With time, the child gets smarter, and before you know it, the student surpasses the teacher.
The quality of the output depends on the quality of the input/prompt. I think most of the specialists were not yet familiar with how to properly "talk" to the chatbot to get the most effective/human-like answers.
True.
However, just as a sidenote, having the quality of the instructions be too dependent on the customer using the right language and asking the right questions while rarely asking the customer questions, is something that I think AI can still improve on.
@@anoukdevries8144 Well it's kind of the big counterargument that speaks for chatGPT. There's a learning curve when using a computer, there's a learning curve when using a phone, Microsoft Word, Zoom, writing a resume, writing an email, doing your taxes, etc. etc. of course there's going to be a learning curve when interacting with LLMs.
I feel like the title should've been "A.I Tries 20 jobs with just one prompt". Realistic, most of those answers could've been enhanced with other prompts.
Right, you have to talk to AI just right. AI must truly comprehend the perspective of the information you are attempting to extract. Better prompts, and AI would undoubtedly have taken some of their jobs lol.
With many of them confidently staring at the camera and saying “AI can’t do my job”, im guessing this video is not gonna age well. 😂
The way some of them try to justify their jobs is hilarious! :D
Yea, at least the comedian does it for a living. Not sure if he intended the last part as a joke or not...
Yeah. They're too confident after checking prompt on a generic and even outdated AI
the fitness trainer's job can be done perfectly by the AI lmao
A thing that gets on my nerves with this video is that they usually say they want more information, but they don’t ask the ai for that information. ChatGPT will elaborate on the information it just provided if you ask it to.
This video will age like fine milk. I would love to see the look on the face of the firefighter when he is replaced by an AI operated drone
@@arthurdurham definitely a net benefit, but there will be fewer firefighters overall. So you're both right.
Once we're at a point where robots can do even the job of a firefighter reliably, I don't think capitalism will exist and neither will jobs.
And why would you want that?
Humans are always going to strongly prefer a human being acting on their behalf rather than a program.
Emergency services, disaster relief, etc are always going to consist mostly of humans being assisted by technology.
AI drones assisting firefighters? 100%.
Replacing firefighters? Lol.
There's also the issue of accountability. You can't hold a black box accountable for bad decisions.
@@jneal4154 you can hold accountable the company that provide you with the drones…
What more did the doctor expect with that type of question?
He had too much copium running through his veins
That's crazy. People really underestimate AI. Everything we learn will be meaningless and we need to have skills that AI can't do
ppl r dum as rocks. they never realize paradigm shifts right in front of them.
Interesting. I think they should have done a follow-up where they had a "prompt engineer" who had some time to research all of these get the chance to come in with each of these people and then show them the best of what the AI can do right now (more or less, obviously the AI doesn't give the same response each time). For example, the therapist one may not have been what I'd want based on what the therapist put in.. but.. someone who knows the system could have had it do a dialogue and zeroed in on stuff.
That is such a good idea
I somewhat agree, but not every person is a prompt engineer. That's the point. So, a "new job" called prompt engineer is invented. The new GPT-4 that can get images, videos, and audios as input will make a huge difference. And these AI tools are meant to help the process. Not overtake. Copilot is a better way to say it.
That defeats the purpose because they're all attempting to simulate what a lay person would do instead of approaching a professional
😎
@@toptrends88 I don't know if prompt engineer makes sense for a job. I think it's up to each professional to figure out how to use the tool better, like with any other tool in their kit. I guess in some cases there could be such a thing as a prompt coach maybe.