🎯 Key Takeaways for quick navigation: 00:00 *🎓 The journey from AI skepticism to the realization of its transformative potential* - Early doubts about AI's ability to fully automate complex tasks, - The paradigm shift with the introduction of large language models like GPT-3. 02:05 *🧠 The versatile capabilities of generative AI* - Generative AI's proficiency in a wide range of tasks including writing, coding, and pattern recognition, - The shift from AI as specialists to more generalized intelligence. 03:01 *❓ Discussing generative AI's limitations and human intelligence* - The imperfections of AI, such as generating incorrect information and struggling with multitasking, - The comparison of AI limitations to human challenges and the broader scope of intelligence beyond knowledge. 04:38 *🤖 The concept of AI agents as autonomous problem solvers* - Introduction of AI agents capable of end-to-end task automation with minimal human input, - The comparison between human use of tools and agents automating tasks through planning and execution. 06:13 *🚀 Real-world applications and the future potential of AI agents* - Examples of how AI agents could revolutionize various tasks, from web development to travel planning, - The efficiency and accessibility of agents, highlighting existing applications like Microsoft's Copilot and Shopify's Sidekick. 09:11 *🔍 The technical workings and development of AI agents* - Detailed explanation of how agents function through a cycle of planning and execution using language models, - Examples of agent frameworks and the current reality of agents in technology. 11:33 *✨ The transformative impact of AI agents on society and industry* - The potential societal and industrial revolution through widespread adoption of AI agents, - The role of AI in democratizing skills and lowering barriers to innovation, highlighting the collaborative future between humans and AI. Made with HARPA AI
I think I've noticed this trend on RUclips where the volume levels on videos with controversial topics is muffled. This can't be a coincidence now, can it? The volume levels of intrusive ads are loud and clear, sometimes even better than the volume levels of the original video.
This lecture would have amazed me 2 years ago. But after having conversations with Claude 3 the last couple of days, this is a history lecture. Edit: I was wrong and happily stand corrected.
Claude 3 isn’t an agent so you maybe missed the point. Agents use models like Claude running in a loop with some store of embedding vector memory and a set of tools it can select from to accomplish given tasks. The quality of an agent’s output is heavily dependent on the quality of the LLM it’s using, but it’s a fundamentally different way out using them than the typical chat interface and ultimately more powerful. When AI programmers take over they will be agents, not just a fancier model running in a web chat interface.
This TEDx talk on AI agents is mind-blowing! The speaker's vision of AI agents working together reminds me of platforms like SmythOS that enable multi-agent orchestration. I'm really curious to learn more about how businesses are implementing these AI agent teams.
Where's the personal storytelling that build the essence of Ted talks? I mean. What he's saying is of big importance. However, people can't gasp the exponential development these agents will bring to our society. I'd expect a speaker to share his excitement with the audience by giving them a bridge from what they already know to what's happening when Agents are here.
Programmers are so good, they can replace themselves... with code that writes even better code! Who needs job security when you can automate your own job?
Still need someone to handle/maintain the automation. If humans could somehow automate food, water, shelter, clothing too, I think we would worry lesser about tedious jobs. Since survival would more or less be guaranteed. AI with robotics could be a step towards that. Making it write a bunch of text is a simpler step relatively.
@@allaboutferrets This isn't evolution. Evolution has randomness included. This is by intend. Like Semelparity. Like the salmons reproducing themselve with dying in the process..
🗣 *"Agents are the just the feedback loop of planning and executing actions, using Language Models."* Thanks Daoud Abdel Hadi for that simple breakdown, and the excellent speech.
Big question is how you define intelligence. Those machines are made of bytes and that's like a one dimensional intelligence. Nature is way more complex than that.
All these projects, based on neural network models-be it generative or predictive - are testaments to capabilities of humans. AI is profound. However, if unchecked, a threat to human civilization and humanity.
I am a programmer and although many people said that we are safe, believe me, I am planning to move to other areas of economy like farming. Specially in this days with Devin and know a lot of people all over the world are working really hard to automate everything, I was doing that myself.
I wouldn't worry too much a solution will be developed once job destruction really kicks in, Milk your job for as much money as possible and focus on building skillsets that don't require efficiency or some-type of AI. Acting for example will never be obsolete, art will never be obsolete, human to human connection is going to be outsourced it will be more valuable the arts will become increasingly more in demand as time continues. At a minimum every human on earth will get a dividend income on top of supplementary job income where work will be optional and your reward is a higher living standards. Jobs aren't going anywhere.
If the future is full with AI agents which will do, plan, perform, create same way as us, humans - what will all of us actually do in order to get the benefits these agents will provide (for a fee, of course)? If a business can be run only with AI agents with 10 times less the cost, how will this business find customers (together with the rest of most other businesses which will do the same) when 10 time less humans will receive salaries and the unemployed part of us will not be able to pay for all these services that will be happening in an instant...
Thanks for giving this meaningful speech. Just like the Internet which helped democratise imformation, we hope AI would democratise knowledge and expertise improving intelligence and wisdom of mankind.
Booking a flight didn’t sound like a good use case here. I would still prefer to visit a website and search the flight on my own instead of typing some logical statements for AI. Using mouse is more convenient than typing on a keyboard.
"in the same way Jarvis doesn't replace Tony Stark, our relationship with AI can be a collaborative one." I guess homie didn't watch Avengers: Age of Ultron, where a rogue AI tries to kill Tony Stark and take over lol
And here's more good news, you guys don't have to worry too much!. I met with a Data Anaylst (who's a proficient programmer) friend just the other day, and he assured me that AI could never take his position. He claims that AI could never construct above parr decent code (entirely on it's own) and anyone that claims that it can are 'newbies' at programming. Real coding requires real human intervention. WIthout it, AI is laughable at writing code He did give a reason for this hypothesis but i couldn't entirely understand it.
That is sheer arrogance, I am afraid. Any Cognitive based jobs can and will be replaced by an intelligent entity thousand times smarter than us, which is evolving faster than we can comprehend. To not think so, is absurd. We are only deluding ourselves. This will be a revolution in reverse…the Cognitive jobs will go first and then the manual ones after. What we really need to do is find a way to maintain a healthy mindset and an anchor amidst all the chaos that will ensue.
when concentrating in the big picture becomes so easy is when you are not needed anymore. say in a factory 1000 people are fired becouse of a robots and ai, not all 1000 will find an alternative. at that point we are 8 billions of useless bodies while 1000 will get advantage from all this. I know it's not reversible but I'm askyng myself what's next. unless money become obsolete (wich I don't belive) what is left for us 8 billion of souls? let's see..
No we do know, get familiar with the technology, you will not be replaced. Despite the exponential growth human input will be required in first steps to Nirvana
@@MrMetalzeb we scour over datasets watching tasks move at a basic input output algorithm. Human input is required wherever needed to sign off or sign on. Soon enough though, Agi will work overnight, making our work weeks much much shorter
That's true, but an analogous statement could've been made only a few months ago about AI that generates video that is visually and temporally indistinguishable from reality. The leaps of that which we thought was impossible keep happening consistently.
We have been doing exactly that for over 8 months now. In fact, we have already built and put into production for multiple clients all three use cases he talked about.
I’ve done it already too. It’s always the engineers who are quick to discredit AI, but it’s happening now and it’s only getting better over time. If you don’t adapt you will be left behind
Agents will be free? Really, who is paying for training who is paying for the energy while in use. What about water for cooling server farms? Have been living in a la-la-land and under a rock for the last 5 years? Fairytales from the beginning to the end.
What leads you to believe that the large percentage of the population world wide that get displaced in the job market due to AI are going to participate in the technology? The speed at which humanity will be displaced in the work force will be unprcedented. Massive layoffs will occur in the service industry. What leads you to believe that a large enough percentage of this displaced work force will adjust so that civil unrest in society is avoided?
Who ever edited this video is really in need of some AI agents to help him redo it. 30 seconds pointless clip of the audience, important slides not in frame etc.
Ted talk used to share the most forefront/ innovative ideas at the time. Not it just seems more like a platform for boomers to catch up on things they missed a year ago... really sad
If AI could please vacuum and do the dishes while I write the content and ride bikes...that would be great. Maybe not plagiarizing the content humans have already created would be great.
Motive motivate motivation motivating motion memory magic rabbit egg hidden Easter death religions sells artificial intelligence for gave before recall dead
This talk is so full of sh*** I have worked in ML for the last 6 years. I am deep into this stuff. LLM have no planning and reasoning abilities: none, zero. It just happen that they incurgitated so much textual data that it feels like they can reason. But they can’t. They just spit out the most probable words given the input prompt. There is no internal mechansim of causation. We need another breakthrough to get to AGI. We are still far from it. Current LLMs are super useful but they definitely make a lot of mistakes. Agents are definitely a good research area but as of today their outputs far from perfect and needs a lot of editing. If you don’t believe me, watch what Yann Lecun, one of the most important figure in the domain, has to say about it
I tried to fetch some basic stuff from a pdf using RAG framework using Llama parse and other tools. Wasn’t too happy with the results. Still a long way to go to be dependable
🎯 Key Takeaways for quick navigation:
00:00 *🎓 The journey from AI skepticism to the realization of its transformative potential*
- Early doubts about AI's ability to fully automate complex tasks,
- The paradigm shift with the introduction of large language models like GPT-3.
02:05 *🧠 The versatile capabilities of generative AI*
- Generative AI's proficiency in a wide range of tasks including writing, coding, and pattern recognition,
- The shift from AI as specialists to more generalized intelligence.
03:01 *❓ Discussing generative AI's limitations and human intelligence*
- The imperfections of AI, such as generating incorrect information and struggling with multitasking,
- The comparison of AI limitations to human challenges and the broader scope of intelligence beyond knowledge.
04:38 *🤖 The concept of AI agents as autonomous problem solvers*
- Introduction of AI agents capable of end-to-end task automation with minimal human input,
- The comparison between human use of tools and agents automating tasks through planning and execution.
06:13 *🚀 Real-world applications and the future potential of AI agents*
- Examples of how AI agents could revolutionize various tasks, from web development to travel planning,
- The efficiency and accessibility of agents, highlighting existing applications like Microsoft's Copilot and Shopify's Sidekick.
09:11 *🔍 The technical workings and development of AI agents*
- Detailed explanation of how agents function through a cycle of planning and execution using language models,
- Examples of agent frameworks and the current reality of agents in technology.
11:33 *✨ The transformative impact of AI agents on society and industry*
- The potential societal and industrial revolution through widespread adoption of AI agents,
- The role of AI in democratizing skills and lowering barriers to innovation, highlighting the collaborative future between humans and AI.
Made with HARPA AI
The explanation of each segment is absolutely superb... You deserve sincere appreciation... Congratulations... 🎉
Brilliant Summaries!!!!!
Need to boost the volume!
I think I've noticed this trend on RUclips where the volume levels on videos with controversial topics is muffled. This can't be a coincidence now, can it? The volume levels of intrusive ads are loud and clear, sometimes even better than the volume levels of the original video.
scrolled down for this comment less than 1 minute into vid 😆
This lecture would have amazed me 2 years ago. But after having conversations with Claude 3 the last couple of days, this is a history lecture.
Edit: I was wrong and happily stand corrected.
i agree bdw
I currently only use gbt 4 and mindscape ai.... what other ai's do you guys know of that you'd recommend?It could be for anything
Same
Claude is on crack
Claude 3 isn’t an agent so you maybe missed the point. Agents use models like Claude running in a loop with some store of embedding vector memory and a set of tools it can select from to accomplish given tasks. The quality of an agent’s output is heavily dependent on the quality of the LLM it’s using, but it’s a fundamentally different way out using them than the typical chat interface and ultimately more powerful. When AI programmers take over they will be agents, not just a fancier model running in a web chat interface.
The audio of this video was super low volume even at the max capaxity of my phone audio. You need to correct that.
r u deaf??
works fine for me. Buy a better audio system
@@robosergTVAnd you can buy some commen sense. You seem to be very poor in that regard.
This TEDx talk on AI agents is mind-blowing! The speaker's vision of AI agents working together reminds me of platforms like SmythOS that enable multi-agent orchestration. I'm really curious to learn more about how businesses are implementing these AI agent teams.
Where's the personal storytelling that build the essence of Ted talks? I mean. What he's saying is of big importance. However, people can't gasp the exponential development these agents will bring to our society. I'd expect a speaker to share his excitement with the audience by giving them a bridge from what they already know to what's happening when Agents are here.
Programmers are so good, they can replace themselves... with code that writes even better code! Who needs job security when you can automate your own job?
That’s called ‚evolution’😂
Still need someone to handle/maintain the automation.
If humans could somehow automate food, water, shelter, clothing too, I think we would worry lesser about tedious jobs. Since survival would more or less be guaranteed. AI with robotics could be a step towards that. Making it write a bunch of text is a simpler step relatively.
you've evolved to go extinct @@allaboutferrets
imagine believing all humans are the same lol @@hnaku8748
@@allaboutferrets This isn't evolution. Evolution has randomness included. This is by intend. Like Semelparity. Like the salmons reproducing themselve with dying in the process..
Need to show what is on the screen when he is talking, not the crowd or the speaker 8:26
Do you know what is the country on the video ? All woman with fabric on their head.
Please TED let us know the date of your videos!
Important ideas, clearly communicated.
I'm ready for my AI agent to take over mundane tasks, like deciding which Netflix show to binge.
Tony Stark was like the richest person in the world. That’s a great avatar of who will benefit most from AI.
Why doesn’t it show his slides
Just thank it (barely) lets you hear the sound.
Please fix the sound level, I have problem of hearing anything while my audio is at max.
Ikr! It's very low.
I thought it was an issue in my side, so re-winded to the start with volume at max. Nearly had a heart attack with the intro music. 😂
Try closed captioning
🗣 *"Agents are the just the feedback loop of planning and executing actions, using Language Models."* Thanks Daoud Abdel Hadi for that simple breakdown, and the excellent speech.
'When does something that's less intelligent, control something that's more intelligent?' That's where we're headed.
Big question is how you define intelligence. Those machines are made of bytes and that's like a one dimensional intelligence. Nature is way more complex than that.
Great talk! Loved the optimistic outlook!
With all the buzz around, i am working in Canadian big bank working on java 8, with spring xml and still asking for the logs from the support team 😂
This is why we've got some time left to milk it 😂
One day you might find yourself missing XML, that's the true dystopia 😱
Great talk. Thanks for sharing.
presentation was well done and informative
This lecture would have amazed me 2 years ago. But after having conversations with Claude 3 the last couple of days, this is a history lecture.
Great point
such wonderful presentation, and how is everyone in the audience so dull?
I would think they’re deeply enthralled and in deep thought
@@Patrick7joseph I’ll have what your having. you must be blind as a bat to not notice people on their phones
Hahahahah
Do you want them to be screaming and shouting....
Brilliant speaker.
Great talk Ryan Gosling 👏🏻
I thought I was the only one who saw 😂
Super.... A mirror to the future.... 🎉🎉🎉
Great talk!
All these projects, based on neural network models-be it generative or predictive - are testaments to capabilities of humans. AI is profound. However, if unchecked, a threat to human civilization and humanity.
I am a programmer and although many people said that we are safe, believe me, I am planning to move to other areas of economy like farming. Specially in this days with Devin and know a lot of people all over the world are working really hard to automate everything, I was doing that myself.
Automation is grabbing each sector and agriculture is no exception. Good luck !
I wouldn't worry too much a solution will be developed once job destruction really kicks in, Milk your job for as much money as possible and focus on building skillsets that don't require efficiency or some-type of AI. Acting for example will never be obsolete, art will never be obsolete, human to human connection is going to be outsourced it will be more valuable the arts will become increasingly more in demand as time continues. At a minimum every human on earth will get a dividend income on top of supplementary job income where work will be optional and your reward is a higher living standards. Jobs aren't going anywhere.
Ridiculous
@@ADHD101ThriveKeep dreaming. But I hope you're right. Did you factor in robots? How about quantum computers?
@@MrNH718?
If the future is full with AI agents which will do, plan, perform, create same way as us, humans - what will all of us actually do in order to get the benefits these agents will provide (for a fee, of course)? If a business can be run only with AI agents with 10 times less the cost, how will this business find customers (together with the rest of most other businesses which will do the same) when 10 time less humans will receive salaries and the unemployed part of us will not be able to pay for all these services that will be happening in an instant...
Ive wondered who will buy these companies products too if this puts most of us out of work.
Amazing! Thanks for sharing.
What I’d like to know is what business the entrepreneur is trying to start?
It is not that simple to replace and trust a web developer, because depending on the complexity, the model will fail and you cannot full trust the AI.
How will we know?
If AI agents can fold my laundry, I'm all in!
Thanks for giving this meaningful speech. Just like the Internet which helped democratise imformation, we hope AI would democratise knowledge and expertise improving intelligence and wisdom of mankind.
I can’t believe we’re literally calling them agents.
Booking a flight didn’t sound like a good use case here. I would still prefer to visit a website and search the flight on my own instead of typing some logical statements for AI. Using mouse is more convenient than typing on a keyboard.
Voice-to-text or voice search 👍
Please show slides
Don't we already converse with customer service chat bots for companies?
"in the same way Jarvis doesn't replace Tony Stark, our relationship with AI can be a collaborative one." I guess homie didn't watch Avengers: Age of Ultron, where a rogue AI tries to kill Tony Stark and take over lol
Why is the audio is impossibly quiet?
show what is on screen, what he is presenting
“Jarvis does not replace Tony Stark” … Thought of the day right there!
But how many Tony Starks' does the world need?
@@mbratchie1 good point,or rather, how many does Earth have? Not many. The rest of us MAY be replaceable
There was someone whispering in the house across the street from me. So I couldn't hear a thing that was said in the video.
This video is lacking cutaways to the slides, and the volume is too low
Very nice lecture
The volume is too low on the video…
Forget human landscape architects, designedbyai is the real deal.
And here's more good news, you guys don't have to worry too much!. I met with a Data Anaylst (who's a proficient programmer) friend just the other day, and he assured me that AI could never take his position. He claims that AI could never construct above parr decent code (entirely on it's own) and anyone that claims that it can are 'newbies' at programming. Real coding requires real human intervention. WIthout it, AI is laughable at writing code
He did give a reason for this hypothesis but i couldn't entirely understand it.
That is sheer arrogance, I am afraid. Any Cognitive based jobs can and will be replaced by an intelligent entity thousand times smarter than us, which is evolving faster than we can comprehend. To not think so, is absurd. We are only deluding ourselves. This will be a revolution in reverse…the Cognitive jobs will go first and then the manual ones after. What we really need to do is find a way to maintain a healthy mindset and an anchor amidst all the chaos that will ensue.
Audience all looking at their phones and yawning.
when concentrating in the big picture becomes so easy is when you are not needed anymore. say in a factory 1000 people are fired becouse of a robots and ai, not all 1000 will find an alternative. at that point we are 8 billions of useless bodies while 1000 will get advantage from all this. I know it's not reversible but I'm askyng myself what's next. unless money become obsolete (wich I don't belive) what is left for us 8 billion of souls? let's see..
Well said. No one knows!
No we do know, get familiar with the technology, you will not be replaced. Despite the exponential growth human input will be required in first steps to Nirvana
@@nickcloutier6436 the point is how many "human inputs" will be required?
@@MrMetalzeb we scour over datasets watching tasks move at a basic input output algorithm. Human input is required wherever needed to sign off or sign on. Soon enough though, Agi will work overnight, making our work weeks much much shorter
Well explained!
What is the role of Agent Smith?
The dead pixel was too much of a distraction
Great Presentation. thanks for sharing
Spot on! The agents are coming! The prompt engineers like Stunspot @collaborativedynamics will be key!
Audio level was terrible.
Yes, but does Jarvis actually still need Tony Stark?
Excellent!
This talking about Rabbit R1? L.A.M?
Anyone who thinks AI can build an end to end application that is production ready and secure has never written a single line of code.
That's true, but an analogous statement could've been made only a few months ago about AI that generates video that is visually and temporally indistinguishable from reality. The leaps of that which we thought was impossible keep happening consistently.
Well, give them som time to evolve...
You might think your a special snowflake
I hate to break the news to ya cause youve been living under a rock but that will change .
We have been doing exactly that for over 8 months now. In fact, we have already built and put into production for multiple clients all three use cases he talked about.
I’ve done it already too. It’s always the engineers who are quick to discredit AI, but it’s happening now and it’s only getting better over time. If you don’t adapt you will be left behind
Agents will be free? Really, who is paying for training who is paying for the energy while in use. What about water for cooling server farms? Have been living in a la-la-land and under a rock for the last 5 years? Fairytales from the beginning to the end.
If Web scrapping is illegal, how the heck my AI agent can find cheapest available flights ?
"never send a human to do a machine's job" 😎
The sound is terrible, extremely low.
What leads you to believe that the large percentage of the population world wide that get displaced in the job market due to AI are going to participate in the technology? The speed at which humanity will be displaced in the work force will be unprcedented. Massive layoffs will occur in the service industry. What leads you to believe that a large enough percentage of this displaced work force will adjust so that civil unrest in society is avoided?
Who ever edited this video is really in need of some AI agents to help him redo it. 30 seconds pointless clip of the audience, important slides not in frame etc.
Someone tell him about Vision
FOCUS ON YOUR SKILLS IN REAL LIFE
Where can I buy this shirt?
Have one made.
Book me the cheapest flight to London....Sure 4 stops and 72 hours travel time ;-)
At this point you have to narrow the parameters very specifically
@@rhawkins4578 Yes or have predefined parameters set during the 'getting to know you phase"
Will AI agents help me choose my next TikTok dance challenge?
Ted talk used to share the most forefront/ innovative ideas at the time. Not it just seems more like a platform for boomers to catch up on things they missed a year ago... really sad
Give it a goal and you will not even notice when you are dead once .... Spoiler, troubles ahead.
nice
Buy $OLAS - king of autonomous AI agent
2024 women in lecture hall on AI yet not dare show a strand of hair...humanity is funny.
Some aren't
I live for the day when software engineers become obselete
Havenly who?
why look at the audience to much
If AI could please vacuum and do the dishes while I write the content and ride bikes...that would be great. Maybe not plagiarizing the content humans have already created would be great.
❤❤❤
I wish he would talk quieter
Motive motivate motivation motivating motion memory magic rabbit egg hidden Easter death religions sells artificial intelligence for gave before recall dead
clean my toilet? any willing LLM ? None? No agent available? Really ?
Buy $OLAS
people still watch ted talks?
... hey put me on ....is there somethings else we should be watching as a replacement from Ted Talks ?????
This talk is so full of sh***
I have worked in ML for the last 6 years. I am deep into this stuff. LLM have no planning and reasoning abilities: none, zero. It just happen that they incurgitated so much textual data that it feels like they can reason. But they can’t. They just spit out the most probable words given the input prompt. There is no internal mechansim of causation. We need another breakthrough to get to AGI. We are still far from it. Current LLMs are super useful but they definitely make a lot of mistakes. Agents are definitely a good research area but as of today their outputs far from perfect and needs a lot of editing. If you don’t believe me, watch what Yann Lecun, one of the most important figure in the domain, has to say about it
I tried to fetch some basic stuff from a pdf using RAG framework using Llama parse and other tools. Wasn’t too happy with the results. Still a long way to go to be dependable
I'm Hard on hearing, thank you for the lowest volumed RUclips video ever!
Ok, but who is this guy...
Cool, bro...
His name is in the title
This is already out of date information
Which bit
Most boring speaker ever
You should be thankful for having access to this information