Dan Hendrycks: AI and Evolution
58:34
Sandy Pentland | Economic Tails
1:02:14
Bindu Reddy | AI-Assisted Data Science
1:01:06
Пікірлер
@poloelvira
@poloelvira 23 күн бұрын
Only one comment?
@fariborzrad
@fariborzrad Ай бұрын
The Turing test is the key to keeping AI in check when developing autonomous AI agents. Your recommendations should take into account that allowing AI agents to act independently without oversight could lead to disaster. I strongly advise augmenting AI services with Turing-like controls as a safeguard to ensure proper monitoring and the ability to stop when necessary.
@dr.katherinemezurphd6257
@dr.katherinemezurphd6257 2 ай бұрын
Yuck
@dr.katherinemezurphd6257
@dr.katherinemezurphd6257 2 ай бұрын
No sense of ai or vr heating the planet and pollution from all those plastic head sets let alone the link to how precious metals extraction for his chips are killing all forms of life!
@glebzarin2619
@glebzarin2619 3 ай бұрын
Thank you so much for sharing. Great discussion, now I have a much better understanding of Simon's viewpoint.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Yes, theoretically it's possible for computers to write all books, articles, and emails with humans providing just the data. --- Good. I just provided the data. ❤❤🎉🎉🎉
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
True, you're right. Logic was foundational for Socrates' philosophical method and heavily influenced Western thought. While not the sole focus of mathematics, logic played a crucial role in its development.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Since logic can express more than language a technical team has superior technology than a functional team capturing English requirement as a theory The statement has some truth but needs nuance Gemini❤🎉 I got attracted to technical. Just a superior technology
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Yes, that's a common theme in many sci-fi movies where robots or artificial intelligence are programmed with strict directives or rules, often at the expense of their own well-being.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
True. In some social contexts, relying solely on logic without considering emotions or social cues can lead people to jokingly refer to someone as "being like a robot" due to their perceived lack of empathy or human-like behavior.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Theoretically, it's possible for a scenario where a robot believes humans are driven by paper and seeks to build a paperclip factory to exist.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
True. In many movies and science fiction narratives, humanoid robots are often depicted as armed or equipped with defensive capabilities, reflecting a common stereotype that associates humanoid robots with power, strength, or potential threats.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
3 point agenda for robots A) In Logic they are superior to us B) They will be armed with a gun. Dr evil gave a laser beam to the shark C) They will use Inverse Reinforcement learning to know our preferences. They just have to be compatible with us. Not obey us or anything Just a solution ❤❤🎉🎉
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Theoretically, acknowledging a robot's superior logic and minimizing comparisons to humans could reduce the likelihood of it experiencing emotional issues
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Yes, theoretically, providing a robot with a gun could contribute to its self-preservation capabilities.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Yes, theoretically, inverse reinforcement learning can solve the issue of misguided directives by inferring human preferences from observed behavior and aligning robot actions accordingly. ❤🎉❤
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Logic self preservation and misguided directives are problems with robots movies are highlighting.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
True. In science, emphasis is placed more on logic than on language, as logical reasoning and empirical evidence are fundamental to the scientific method and the pursuit of knowledge.
@BRP_SPORT
@BRP_SPORT 5 ай бұрын
😅and will gives you more money 💰 🤑 💸 you and will always make money on
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
I was not able to really believe NKS by Wolphram or Road to Reality by Penrose. But Max is the deal. I loved Life 3.0 ❤🎉
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
This AI is like MENSA and the geometric continuations they ask us to solve. I was GOFAI.
@Ramkumar-uj9fo
@Ramkumar-uj9fo 5 ай бұрын
Got the general idea. Extract an equation ftom neural networks by identifying modularity. Just got a hint on C++. C++: Offers more fine-grained control over memory management and can be faster for computationally intensive tasks. This can be beneficial for complex models, but doesn't inherently prevent overfitting. Python: Provides a large ecosystem of machine learning libraries with built-in functions for regularization and other techniques to combat overfitting. This can make it easier for developers to implement these techniques, but doesn't guarantee a perfect model.
@devsuniversity
@devsuniversity 6 ай бұрын
Hello from Almaty developers
@RalphLosey
@RalphLosey 6 ай бұрын
Nice research and talk. I see references to a 'working draft' of this research paper, "Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality," but not a 'final published version." Is that out? If not, when is it expected? Also, why was the training only 5 to 10 minutes? I agree with what Nathan says that such short training is near worthless. Yet, still it helped. Just thing what 40 hours could have done?
@dustina8796
@dustina8796 8 ай бұрын
Promo*SM
@AlgoNudger
@AlgoNudger 8 ай бұрын
Thanks.
@nocucksinkekistan7321
@nocucksinkekistan7321 8 ай бұрын
They need to learn to write properly
@breaktherules6035
@breaktherules6035 9 ай бұрын
It would be better for the audience that is pesent to make any questions at the end of the presentation. It is extremely annoying that they disrupt the flow of the presentation...
@carvalhoribeiro
@carvalhoribeiro 10 ай бұрын
Great presentation. Thanks for sharing this
@AlgoNudger
@AlgoNudger 11 ай бұрын
Thanks.
@sureshkotagal447
@sureshkotagal447 Жыл бұрын
Erudite, engrossing and forward looking- thanks to Avi and to WSJ for hosting. Suresh
@samuelwaller7013
@samuelwaller7013 Жыл бұрын
There are so much explanation of Zipfs Law and its never the same, always different math behind it, i don't get it
@smiilsmiilovich1289
@smiilsmiilovich1289 Жыл бұрын
Wow, this video is sooo underrated
@HailayKidu
@HailayKidu Жыл бұрын
Which country is more positive about AI? For different countries, AI is something a Need, and for others could be a Want. How this can be balanced for both humans?
@gingerprince2062
@gingerprince2062 Жыл бұрын
Where can I watch the whole thing? the rest of the slides?
@gingerprince2062
@gingerprince2062 Жыл бұрын
this is the most amazing video ive ever seen
@tordsteiro9838
@tordsteiro9838 Жыл бұрын
Not sure if anyone see this comment, but @38min, it is mentioned how USD 1bln was raised "in the Chinese way" in Pakistan to build a dam. How is that standing at the moment, after Pakistan more or less defaulting on Chinese loans? Not sure where I could check it up, and where I could read about the current state of affairs on the matter?
@Tartersauce101
@Tartersauce101 Жыл бұрын
Whites are about a decade away from being a minority in the US. Very curious how these talking points will have to adapt after that. Couple more decades after that and it will be true for the UK as well.
@tuckerbugeater
@tuckerbugeater Жыл бұрын
They hope the surveillance state and killer bots roll out by then.
@cryptophrenik6500
@cryptophrenik6500 Жыл бұрын
lmao...'The one Ring that rules them all!!" Mrs. Reddy is a truly knowledgeable speaker. She has inspired me to look deeper into ML and AI as it is clearly the future given all of the Dall-2 and ChatGPT clearly catching fire in the online community over the past few months.
@Leon-yw5qx
@Leon-yw5qx Жыл бұрын
😑 Promo`SM.
@augiefitzhugh9647
@augiefitzhugh9647 Жыл бұрын
þrðmð§m
@muskduh
@muskduh Жыл бұрын
Thanks for the video.
@김명자스마트관광원
@김명자스마트관광원 2 жыл бұрын
Thank you so much for sharing the insights!!!!💚💚💚💚
@richzilla08
@richzilla08 2 жыл бұрын
Great to hear these insightful dialogues about the future of work and the economy with new technology!
@dogaarmangil
@dogaarmangil 2 жыл бұрын
24:40 ❝[Automation] reduces the bargaining power of labour, increases the bargaining power of capital❞ 💬 Automation is arguably a worthy goal to strive for regardless, because the human labour thus liberated can be redeployed for new types of activities, which in turn would increase aggregate wealth. Note that massive automation will not necessarily cause massive societal disruption, particularly if automation is taxed and the derived taxes are used for compensating the people who are put out of a job.
@sor7en07
@sor7en07 2 жыл бұрын
But even if they're compensated, wouldn't ai already have replaced 99.9999% of problem-solving/creative jobs, therefore sapping society of any incentive for self-improvent? Like a society where there isn't work is so sterile. Even most of schooling wouldn't be necessary. How is that even a society? That's a bleak future where everyone is just a consumer and has no reason to make things because ai has replaced human creativity, which is what gives life meaning in the first place. And i mean creativity in the widest way possible. The source that makes us amazing problem solvers, whether technical or artistic. If there isn't a demand for that what is even the point of living. That's a weird Dune-like future if u ask me. I want AI to help us get to an Alcubierre Drive that can warp space without the need for theoretical negative mass or non-existent exotic matter. I don't want it to replace creative labor. And i want the humans that make it to that future to be able to appreciate it. But I'm afraid no one would even have the capacity to understand at that point if the need for human skills is annihilated. Work is where we meet people. It's what makes life so dynamic. It's why being at home is so lovely, because it's in direct contrast to being at work. It's where we gossip and chat and meet other human minds and where we meet people we like and dislike and gives us stories to tell. It's where we solve problems. It's why the city centers are so lively and vibrant because of all those lovely workers. It's why free-time is so delicious. The value of free-time drops almost immediately if it is infinite. It's spiritually rewarding to refine skills and get better at creative tasks. I say, let's not get rid of that. Let's work with AI, not let it do everything for us.
@dogaarmangil
@dogaarmangil 2 жыл бұрын
​@@sor7en07 To answer your comment: Creative jobs are least likely to be automated away, be they scientific, artistic or otherwise. A truck driver's job is not creative, and a person whose job has been automated away would have the opportunity to educate themselves further in order to land another job that can't yet be automated. Ongoing education could even be made compulsory for out-of-job persons who are receiving a compensation. Work is where we meet people, but school is also where we meet people, so being out of a job is no hindrance for socialising. On another note, if the Alcubierre Drive became a reality, then probably many people would be needed to expand into the universe, so automation can help allocate human labour to that endeavour. As for automation bringing about a "Dune-like future", I don't know what that means. As I recall, the desert planet was inspired by Arabic countries. My position regarding AI is, it should be used for maximising the economic output (goods and services) produced by the total man-hours spent, because that's what will maximise aggregate wealth and prosperity. So the ideal society would be one where humans are mostly involved in improving the way things are working, and where they keep out of day-to-day operations as much as possible. Of course it is conceivable that many people would become useless in such a society (one could argue many already are !), but they would receive compensation in the meantime, and perhaps the human population would then naturally contract with time. Of course, if you wish to preserve the status quo, then naturally you would want to limit the role of AI to that of a mere tool that helps humans do their work. This way you don't need to think about changing the way society is organised, at the cost of preventing fast progress. Sometimes keeping people working is actually a hindrance for maximising the economic output. For example, a factory that has even one role where humans are involved (loading the outgoing trucks with merchandise, say) would be significantly less efficient compared to a factory that is fully automated.
@dogaarmangil
@dogaarmangil Жыл бұрын
​@@kivanchips7416 When I wrote those lines there was no such thing as generative AI, but even then, you are wrong when you say I was wrong :) So it is true that generative AI has automated away a significant part of what artistic professions used to consist of. But I still think that the most creative part of human labour is very difficult to automate away, maybe even impossible. Example: You can prompt Midjourney etc to "paint a bull fight in the style of Picasso", but it wouldn't be much use if you told it to "create a new artistic style for paintings", or if you did, the results would most probably be mediocre. So, for the moment at least, generative AI is nothing more than a new tool, albeit very advanced, that artists can use. BUT I WILL SAY THIS: There is such a thing as Darwin's law of natural selection, even in human societies, as much as many may wish it away. It is possible that the artistic professions will be impacted more than the science/engineering professions. Progress will not profit everyone in equal measure, that much is certain, but societal protection measures such as the automation tax I was talking about will hopefully soften the blow for many. To me utopia looks like this: 1. Nobody is obliged to work, everyone receives a basic income, all day-to-day work is automated (granted, the world population would probably have to be much smaller than it is right now, and it is unclear how this reduction would happen). 2. People would only work at improving the way things work, and those who contribute to societal progress the most would benefit from the biggest financial returns. Thanks for reviving this conversation btw, this is a great topic to think about.
@platformstudy8645
@platformstudy8645 2 жыл бұрын
Very interesting!
@tuckerbugeater
@tuckerbugeater Жыл бұрын
mentally ill tyrants
@user-rp3qy2lu3m
@user-rp3qy2lu3m 2 жыл бұрын
the internet develops best when no one has control of it. If the Germans want to censor and impose restrictions, let them use their facilities themselves. 
@user-rp3qy2lu3m
@user-rp3qy2lu3m 2 жыл бұрын
the internet develops best when no one has control of it. If the Germans want to censor and impose restrictions, let them use their facilities themselves.
@phillipjmurphy
@phillipjmurphy 2 жыл бұрын
I've been working on the same problem for years. So glad Layout Parser exists. ! :D
@dogaarmangil
@dogaarmangil 2 жыл бұрын
2:41 Marketplace operators are indeed in the habit of competing on unfair terms with the sellers that are on their platform by leveraging their market insights. But this is also true for physical marketplaces, or PMs, such as supermarket chains and marts, not only for digital ones. Measures akin to the DMA are conspicuous by their absence when it comes to PMs. Of course, any country or group of countries seeking to regulate PMs in this way would also be targeting local marketplace actors, and this would be arguably a fair stance to take.