Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

DeepMind: A Generalist Agent

deepmind.com

Post reply on HN
Print view View original

241–250 of 345 posts

Re: DeepMind: A Generalist Agent

#241

Post by IIAOPSW »

viksit wrote: post #182

(Former AI researcher / founder here) It always surprises me at the ease at which people jump on a) imminent AGI and b) human extinction in the face of AGI. Would love for someone to correct me / add information here to the contrary. Generalist here just refers to a "multi-faceted agent" vs "General" like AGI. For a) - I see 2 main blockers, 1) A way to build second/third order reasoning systems that rely on intuitio…

>2) Deterministic reasoning towards outcomes. Most statistical models rely on "predicting" outputs, but I've seen very little work where the "end state" is coded into a model. Eg: a chatbot knowing that the right answer is "ordering a part from amazon" and guiding users towards it, and knowing how well its progressing to generate relevant outputs.

Here's a real generalist question? What's the point of conversation? In the sense that its a "social game", what are the strategy sets on a given turn, and what does it even mean to "win"? Forget about Artificial Bullshiters vs Artificial Idea-what-to-say. How can we even speculate if ABS or "real AI" solves the problem better when we don't really specify the problem space or how to recognize a solution?

In terms of calls and appropriate responses and responses to responses and terminal states, what even is conversation at the protocol level?

Re: DeepMind: A Generalist Agent

#242

Post by webmaven »

gonehome wrote: post #27
axg11 wrote: post #22

Slowly but surely we're moving towards general AI. There is a marked split across general society and even ML/AI specialists between those who think that we can achieve AGI using current methods and those who dismiss the possibility. This has always been the case, but what is remarkable about today's environment is that researchers keep making progress contrary to the doubter's predictions. Each time this happens, th…

The closer we get, the more alarming the alignment problem becomes. https://intelligence.org/2017/10/13/fire-alarm/ Even people like Eric Schmidt seem to downplay it (in a recent podcast with Sam Harris) - just saying “smart people will turn it off”. If it thinks faster than us and has goals not aligned with us this is unlikely to be possible. If we’re lucky building it will have some easier to limit constraint like…

> People seem to model it like a smart person rather than something that thinks truly magnitudes faster than us.

Even if you model it like a smart person that doesn't think faster on average, there is the issue that a few minutes later you are dealing with a small army of smart people that are perfectly aligned with each other, and are capable of tricks like varying their clock rates based on cost, sharing data/info/knowledge directly, separation of training and inference onto different hardware, on-demand multitasking without task-switching costs, ability to generally trade-off computational space for time for energy, etc.

A silicon intelligence that is approximately human equivalent gets a lot of potentially game-changing capabilities simply by virtue of it's substrate and the attendant infrastructure.

Re: DeepMind: A Generalist Agent

#243

Post by blueblob »

gurkendoktor wrote: post #111
hans1729 wrote: post #5

I’m not sure how to word my excitement about the progress we see in AI research in the last years. If you haven’t read it, give Tim Urbans classic piece a slice of your attention: https://waitbutwhy.com/2015/01/artificial-intelligence-revol... It’s a very entertaining read from a couple of years ago (I think I’ve read it in 2017), and man, have things happened in the field since then. If feels like things truly start…

Not you specifically, but I honestly don't understand how positive many in this community (or really anyone at all) can be about these news. Tim Urban's article explicitly touches on the risk of human extinction, not to mention all the smaller-scale risks from weaponized AI. Have we made any progress on preventing this? Or is HN mostly happy with deprecating humanity because our replacement has more teraflops? Even t…

I feel exactly the opposite. AI has not yet posed any significant threats to humanity other than issues with the way people choose to use it (tracking citizens, violating privacy, etc.).

So far, we have task-driven AI/ML. It solves a problem you tell it to solve. Then you, as the engineer, need to make sure it solves the problem correctly enough for you. So it really still seems like it would be a human failing if something went wrong.

So I'm wondering why there is so much concern that AI is going to destroy humanity. Is the theoretical AI that's going to do this even going to have the actuators to do so?

Philosophically, I don't have an issue with the debate, but the "AI will destroy the world" side doesn't seem to have any tangible evidence. It seems to me that people seem to take it as a given that it's possible AI could eliminate all of humanity and they do not support that argument in the least. From my perspective, it appears to be fearmongering because people watched and believed Terminator. It appears uniquely out-of-touch.

Re: DeepMind: A Generalist Agent

#244

Post by liuliu »

viksit wrote: post #182

(Former AI researcher / founder here) It always surprises me at the ease at which people jump on a) imminent AGI and b) human extinction in the face of AGI. Would love for someone to correct me / add information here to the contrary. Generalist here just refers to a "multi-faceted agent" vs "General" like AGI. For a) - I see 2 main blockers, 1) A way to build second/third order reasoning systems that rely on intuitio…

For a).1). you basically argued in favor of scalability. We don't know if we are on a training set more or less than a typical human at this point. If I would guess, I think high-bandwidth networked computers can be much more efficient at gathering training set than a single human.

For a).2). you argued in favor of symbolic reasoning. My personal interpretation of that line of thinking is: it helps for us to understand complex thinking machines, but it is not necessarily the building block for the thinking machine.

In the area of AI, there are a lot of opinions, but at the end of the day, builders win the argument. So far, what builders have shown gives me optimistic hope.

Re: DeepMind: A Generalist Agent

#245

Post by webmaven »

gurkendoktor wrote: post #165
gonehome wrote: post #150

Earlier quoted context omitted.

Yeah, I'm not arguing alignment is not possible - but that we don't know how to do it and it's really important that we figure it out before we figure out AGI (which seems unlikely). The ant example is just to try to illustrate the spectrum of intelligence in a way more people may understand (rather than just thinking of smart person and dumb person as the entirety of the spectrum). In the case of a true self-improvi…

One avenue of hope for successful AI alignment that I've read somewhere is that we don't need most laypeople to understand the risks of it going wrong, because for once the most powerful people on this planet have incentives that are aligned with ours. (Not like global warming, where you can buy your way out of the mess.) I really hope someone with very deep pockets will find a way to steer the ship more towards AI s…

> I really hope someone with very deep pockets will find a way to steer the ship more towards AI safety. It's frustrating to see someone like Elon Musk, who was publicly worried about this very specific issue a few years ago, waste his time and money on buying Twitter.

It has occurred to me that social networks are a vulnerable channel which we've already seen APTs exploit to shift human behavior. It's possible that Musk is motivated to close this backdoor into human society. That would also be consistent with statements he's made about "authenticating all real humans."

Re: DeepMind: A Generalist Agent

#246

Post by nonbirithm »

viksit wrote: post #182

(Former AI researcher / founder here) It always surprises me at the ease at which people jump on a) imminent AGI and b) human extinction in the face of AGI. Would love for someone to correct me / add information here to the contrary. Generalist here just refers to a "multi-faceted agent" vs "General" like AGI. For a) - I see 2 main blockers, 1) A way to build second/third order reasoning systems that rely on intuitio…

Personally I'm more worried about what effects sophisticated "dumb" AI will have on human culture and expectations. It would have been hard to imagine we'd be living in this world with our attention assailed by platforms available anywhere when Jobs first held up the iPhone to the world. Just the same, I am not sure that the impending cultural effects from advancements in AI are fully understood at this moment.

For example, I wonder what expectations we will set for ourselves when we pick up a pencil once it becomes common knowledge that production-ready digital art can be created by an AI within minutes. What will people be saying about "old art" in those days? What will we think of people that deliberately choose to disavow AI art?

Re: DeepMind: A Generalist Agent

#247

Post by joe_the_user »

sinenomine wrote: post #211
justinpombrio wrote: post #208

Earlier quoted context omitted.

I am quite scared of human extinction in the face of AGI. I certainly didn't jump on it, though! I was gradually convinced by the arguments that Yudkowsky makes in "Rationality: from AI to Zombies" ( https://www.readthesequences.com/ ). Unfortunately they don't fit easily into an internet comment. Some of the points that stood out to me, though: - We are social animals, and take for granted that, all else being equal…

The human extinction due to would be "hard takeoff" of an AGI should be understood as a thought experiment, conceived in a specific age when the current connectionist paradigm wasn't yet mainstream. The AI crisis was expected to come from some kind of "hard universal algorithmic artificial intelligence", for example AIXItl undergoing a very specific process of runaway self-optimization. Current-generation systems aka…

We should think more about the Human alignment problem.

Absolutely this

The possibility of a thing being intentionally engineered by some humans to do things considered highly malevolent by other humans seems extremely likely and has actually been common through history.

The possibility of a thing just randomly acquiring an intention humans don't like and then doing things humans don't like is pretty hypothetical and it seems strictly less like than the first possibility.

Re: DeepMind: A Generalist Agent

#248

Post by narrator »

viksit wrote: post #182

(Former AI researcher / founder here) It always surprises me at the ease at which people jump on a) imminent AGI and b) human extinction in the face of AGI. Would love for someone to correct me / add information here to the contrary. Generalist here just refers to a "multi-faceted agent" vs "General" like AGI. For a) - I see 2 main blockers, 1) A way to build second/third order reasoning systems that rely on intuitio…

I think the "AGI wants to kill us" meme is just to get us ready for the moment when the authorities unleash the killer robots on us because there's too many of us on the planet. "Whoops, those robots did it all by themselves because it's just inevitable that AGI was going to do that. Haven't you watched a sci-fi movie in the last 50 years?"

Re: DeepMind: A Generalist Agent

#249

Post by webmaven »

TaupeRanger wrote: post #216
adamsmith143 wrote: post #120

Earlier quoted context omitted.

There's no evidence otherwise. You have to believe that the mind has a materialist basis or else you believe in woo woo magic.

There's evidence everywhere, every second of every day. It doesn't follow from the mind having a material basis that it is doing linear algebra calculations like a Python machine learning program. That's quite a leap.

> There's evidence everywhere, every second of every day. It doesn't follow from the mind having a material basis that it is doing linear algebra calculations like a Python machine learning program. That's quite a leap.

Not literally, no. But it is entirely possible that what the human mind is doing is equivalent (in the Turing machine sense), if far more efficient (calorically) and robust (though haphazard and stochastic).

Re: DeepMind: A Generalist Agent

#250

Post by esjeon »

Hellicio wrote: post #146
mhitza wrote: post #50

Earlier quoted context omitted.

I'm skeptical because we are building black boxes. How do you fix something you can't reason about? These billion parameter boxes are outside the reach of your everyday developers. In terms of cost of propping up the infrastructure makes them tenable only for megacorps. Most of us aren't moving goal posts, but are very much skeptic at the things we are being oversold on. I personally think we are still far away from…

They are blackboxes for the normal user the same way as a smartphone is a blackbox. Non of my close family understands the technical detail from bits to an image. There are also plenty of expert systems were plenty of developers see them as blackboxes. Even normal databases and query optimizations are often enough blackboxes. As long as those systems perform better as existing systems, thats fine by me. Take auto pil…

> They are blackboxes for the normal user the same way as a smartphone is a blackbox.

You can't take that approach. The current NN techniques are blackbox-by-nature, and are blockbox to everyone including the devs. Proprietary software is only blackbox to consumers, and large complex software still have insides that can be observed when things go wrong. For NNs, nothing can describe how exactly they work, and each network has to be reverse engineered independently, which is, AFAIK, a separate research field.

> I can't drive 10 Million KM in my lifetime (i think). The cars from Google and Tesla already did.

The length (nor the amount of data) alone doesn't decide the quality of AI. Actually, ROI diminishes rather quickly in the early stage of development. The rest is about picking up corner cases. They can drive 1 parsec, and still would not perform better than the ones we have now.

Also, again, because NN is a complete blackbox, even the devs can't be sure if those corner cases are properly reflected to a newly trained network, nor if the new training data didn't impact the performance in other corners. We just don't know for sure. We just take chances. That's the limitation.

Post reply on HN