The Costs of AI and What AGI Would Look Like
HTML-код
- Опубликовано: 11 сен 2024
- #ai #generativeai #llm #aimodels
This video is about
- How much does it costs us to develop AI technologies?
- Where is this AI trend going?
- What AGI would look like and how would we know that it is here?
- What would it mean for us, people, to have AGI?
On this channel, I talk about AI development in terms of new tools, the practical usage of Large Language Models (LLMs), and the latest updates in the field. AI is here to stay and it is better if we learn how to use it to optimize our work. I try to make videos with relatable examples so that anybody can learn to use the power of artificial intelligence in their daily tasks.
🤓 Articles
Microsoft's water consumption: fortune.com/20...
OpenAI Planning for AGI and beyond: openai.com/blo...
What OpenAI really wants: www.wired.com/...
There is quite a lot to unpack in this video but here are a few things that I find interesting.
1. Water and environment. It remains wholly unclear what is meant by "water consumption" in order to evaluate its impact on the social, environmental, politico-economical or other systems (even assuming we can disentangle those). The articles are written in a way that presents this consumption as a black hole - the end point for this water. But if I understood it correctly this water is primarily used as a coolant (when the fresh Iowa air cannot keep up) and while there will be some inevitable loss of water (leaks, evaporation, etc.), most of it will be usable afterwards as there is no reason for it to be polluted (after all it is not cooling nuclear fuel). This should not affect agriculture (following the connection you made) in any significant manner as it can still be used for that. I think that if an AGI was concerned with food production it will just make us obligatory vegetarian/vegan (and continue with the synthetically grown meat) - the vast majority of agrarian land on the planet is used for cultures to feed cattle for meat production rather than produce directly for human consumption. In any case, the production of the hardware needed for these centers is degrading the environment with much more devastating impact when it comes to water pollution. Mining, ore refinement, metallurgy, factories, logistic networks, etc. - AI has increased that but we are pretty happy with all electronic gadgets (phones laptops etc) and the internet which are already the biggest pollution generators, That said, please don't build things that require temperature regulation to dissipate heat in a fucking desert (Arizona.. come on) - that breaks the idiot-meter (unless other factors trump the water requirements).
2. The artificial babies will not be for organ transplants - we can just grow the organ and avoid the ethical hassle. All reasons that come to mind right now are much more grim and the most optimistic one is the beta version of Huxley's brave new world (it is getting close to 100 year old so a good tribute :D) - at the very least you can circumvent any legislative and ethical concerns about extreme experimentation with genetic engineering/modification. However, the question whether we should do something just because we can is an important one and it is likely more important for the development of AGI and the answer has been given - even though we have no idea what that will result in we are going for it. The question whether we should have went for it will be answered by future historians or the lack thereof I guess xD
This comment is getting too long so I will come back with part two focusing a bit more on "curiosity or other motivation" as I do not agree with your position (it might play some role but not a driving force) and the whole notion of the instrumentalist view (technology as a tool we use) of AGI and probably some additional thoughts on the first two points of interest discussed here.
WOW, I don't know if I could resoond to all this but thank you for the comment and feedback ^^
I hope they think like you about the water and try to use it in a sustainable way..
I liked that: "please don't build things that require temperature regulation to dissipate heat in a fucking desert" :D
Yeah, I read a bit on the artificial embryos and I couldn't quite understand their argument as for WHY we are doing this... I mean... okay, testing new meds on people is actually more beneficial than on animals because we get more accurate results but still..
About your last point - okay, I'll wait for the other comment, but yeah.. I've seen a lot of rich people in the US, they aren't interested in more money they're interested in more excitement.. But I guess we could agree that AGI is more than money. Whether it's the curiosity or something else, Idk but when you dont know what you are creating and when money doesn't seem to be important in the contract either, the motivation is somewhere else. [still curious to hear your POV tho!]
I always wonder why you have so few views and even subscribers on your channel. I appreciate your deep thoughts on AI. Today's video is a prime example of this. Thanks
I started Harvard’s course on AI and they talk about how those AI algorithms work. I don’t do myself a favour by having so many diverse videos. It confuses the algorithm. I need to niche down but I can’t decide (yet) what to give up 😄
I always appreciate your encouragement and positive feedback tho 😊✨
Well you came up in my feed, and your video was one of the best I've seen on the subject (and I watch a lot)...so please don't change too much 😉.
@@HelmutBemboka I think I'm gonna make a poll on my community tab and let people vote :D Cos my ADHD brain prefers variety but I know it's not the best way to go about this :)
@@HelmutBemboka ah, and I thank you! :))
@@TeamUpWithAI It may be that your videos are sophisticated, but be sure, these few people who are watching them, love them
Omg the algorithm recommended me this video and I'm GLAD I watched.
This must be the most complete, informative and thought provoking video I've seen lately regarding AIs/AGIs.
Seriously, I can't stress enough how good this was. Hope your channel grows a lot!
Glad you enjoyed it! Thanks for the encouragement too 😊
And yeah, I hope it grows too ;D I make different videos related to AI and I should be a bit more “niched down” buuuuttt.. I’m interested in different things and that doesn’t help ;D
Yeah all of that computation takes energy and cooling like you said, which already directly harms the environment. More than that, as we are becoming more and more dependent on this technology I think the supply risks are also worth mentioning. A huge part of the rare earth elements and semiconductors used in the manufacturing of chips and equipment used in AI development is exported by countries that are rivals in many ways, such as China, Russia, Taiwan, Japan, South Korea, Turkey and the United States. Between the geopolitics of the day and the typical arms race that occurs when any new technology hits the scene, all it takes is an embargo from one of the major players to disrupt everything that we've built on top of AI. An example of this is the 1973 oil crisis.
And speaking of arms races, for me that description also fits the motivation of the powerful people who are so keen on developing more and more sophisticated AI. I agree that curiosity must play a big part, but I think they are also working very hard on positioning themselves - the best way to end up prosperous in an AI-driven world is if you're the one driving AI. Or that would be their logic anyway. I'm really curious how much thought these same people have given to the risks and long-term consequences not just for them, but for everyone else, not to mention the ecological, ethical, even metaphysical implications if you will. This really reminds me of the obsession with space colonization, of the flavor that Elon Musk propagates - "let's just get there, then we'll figure out what to do".
These technological advances are incredibly exciting and I also want to go to Mars on a spaceship piloted by my trusty AI assistant (or am I rather their assistant in that scenario?), however applying a new technology on a global scale before we understand the implications is simply irresponsible. We have many examples of how this can go wrong - from leaded gasoline, to CFCs that degraded the ozone layer, to carcinogenic herbicides on crops all over the world. We thump our chests about how much computational power we have now - why is that not being used to simulate the effects of a technology like AI permeating a globalized world economy? In my opinion the true cost of this is the pressure it puts on an already strained social fabric, where we already struggle to integrate modern technology into our understanding of the world, into our philosophies, ethics and just daily existence.
Thank you for the thoughtful response and sharing your opinion with us! :)
My opinion, a bit negative I know, is that those people don't think about long-term consequences the same way because they don't live in the same world as we do. During COVID, while many people lost their jobs, others made more money for example. I agree with you tho, it's about curiosity yes, but also power. I guess going to sleep knowing that 'you made it' is a familiar concept to all of us, but the criteria for "I made it" is quite different.. :D
👁
AGI will be released in 2024!!!
Idk about that. I was reading an article recently about "AGI being canceled" - it waas about all the restrictions imposed on the big tech companies and the pressure to slow down the AI development. I'm not saying I believe everything I read, but I wouldn't expect to see AGI next year. But who knows?! :)
@@TeamUpWithAI Because its hard to get a permanent high paying job these days,life is hard rn😔
ai creates stpid humas that will make it in a diferent version on new planet after time passes.
Our standard education system often creates stupid humans too, so I’m curious what AI would do and create :)