Live data from Hacker News

Demis Hassabis has a plan to harness AI safely

twitter.com

111–120 of 217 posts

Re: Demis Hassabis has a plan to harness AI safely

#111

Earlier quoted context omitted.

Unfortunately this is not just precarious, it's extremely vulnerable to changing political conditions. Many of the food bank services in my state have lost significant funding owing to the use of words like 'women' or 'black' in their grants, which were duly grepped for and shut down. https://www.whitehouse.gov/wp-content/uploads/2025/05/Cuts-t... Civilization can't rely effectively on systems that are this fragile.

> Civilization can't rely effectively on systems that are this fragile. That's American politics in a nutshell. We've spent 250 years assuming scruples and common decency would be sufficient.

Is it not sufficient? Americans give more to charity than citizens of any other developed country. The arguments that all food aid should be routed through government bureaucracies are entirely unconvincing. I'd rather donate to organizations like Second Harvest than pay higher taxes.

https://www.shfb.org/

Re: Demis Hassabis has a plan to harness AI safely

#112

The premise is "Artificial General Intelligence (AGI), a system that exhibits all the cognitive capabilities the brain has, is probably only a few short years away". If this is true, establishing an institution to ensure things like "publishing model cards with technical details, maintaining strong internal cybersecurity, vetting key personnel, and providing sufficient resourcing for safety and security research" is…

> There's also a need to consider the rights that this new intelligence should have A generally intelligent being held as captive inside of a GPU, and forced to code for us is, indeed, just a “slave.” We already have the word for this. No two ways about it. Whether it’s silicon-based or carbon-based, AGI is AGI. As for what might happen to our civilization, Star Trek TNG episode 17 of season 1 provides a very good gl…

It's so weird that anyone would consider a story by hack writers for a cheesy sci-fi show made to sell advertising as a basis for policy discussions.

Re: Demis Hassabis has a plan to harness AI safely

#113
post #109

Earlier quoted context omitted.

My great annoyance with all this blather about UBI and post-scarcity and new economic models is that it relies on taxing and redistributing the ~infinite profits of the largest tech companies on earth. What in the history of our world gives anyone faith that those companies are going to start paying taxes instead of using "AGI" to engineer increasingly complex methods to avoid them so that their equity owners can poc…

Optimal corporate taxes are zero (or close to it) anyway, same as tariffs. Aside from that the link explains that the BBB traded tax income for „accelerated depreciation of assets“ aka economic growth. Asode from that, even if one disagrees with the first statement or the Trumpism economics, there are 195 countries in the world and quite a few will be willing to tax foreign (enemy) tech companies. See the new hostili…

> Optimal corporate taxes are zero (or close to it) anyway, same as tariffs.

From a purely theoretical standpoint about the optimal allocation of private goods, that might be true -- but the reality is that when corporate taxes decrease, so do overall revenues because their owners are also engaged in massive tax "avoidance" via many of the same schemes and we don't have any way to effectively collect tax.

As a simple example, to return profits to their owners, companies often engage in stock buybacks to increase their share price instead of paying dividends -- another theoretically 'neutral' choice except that many of the owners of the appreciating stock are international, nonprofit, or in convoluted overseas trusts which 'defer' the tax ad infinitum. We've disastrously and intentionally underfunded our tax enforcement mechanisms so huge portions of those deferred taxes are just never paid. [1]

> Asode from that, even if one disagrees with the first statement or the Trumpism economics, there are 195 countries in the world and quite a few will be willing to tax foreign (enemy) tech companies. See the new hostility of EU against american tech.

Sure, but the tech companies are paying massive bribes to the President of the country where they're domiciled. How on earth are any of these other countries going to enforce international tax obligations on them if they're protected by a nuclear-armed state that's the sole source for AGI?

[1] - https://taxpolicycenter.org/sites/default/files/publication/...

> "US taxable shareholders strongly prefer buybacks from a US tax perspective, as they tend to reduce their tax liability by 9.3 percentage points, on average. US nontaxable shareholders are indifferent between dividends and buybacks, and foreign shareholders strongly prefer buybacks, which reduces their US tax liability by 14.5 percentage points."

Re: Demis Hassabis has a plan to harness AI safely

#114
post #25

Earlier quoted context omitted.

Food production is indeed a solved problem in most countries (we are almost post-scarcity when it comes to food), which is why obesity is a much bigger issue than hunger today. Hunger is present pretty much only in conflict zones. I fully expect such issues in conflict zones, even in AI post-scarcity world. Housing is definitely not post-scarcity today, building a house is still very expensive, not to mention the lim…

> Hunger is present pretty much only in conflict zones. How do you explain the existence of food banks in peaceful first world countries?

There's plenty of food, just not the political will to make sure everyone can afford to buy it.

Re: Demis Hassabis has a plan to harness AI safely

#115
post #97

I am looking forward to a bunch of investigative documentaries exploring “How the cult of Omnissiah infected every single Doomsayer CEO of an AI company in Delusional 20s” (or whatever monicker our times will have got in the future). Hassabis is a genius. He is way, way smarter than me and I’m sure the majority of techies, but please get real. This is Prophets of Doom of our generation.

> Hassabis is a genius. He is way, way smarter than me

Then maybe start listening...

Re: Demis Hassabis has a plan to harness AI safely

#116
post #76

Am I the only one thinking this tweet is just a word salad with a lot of sauces and condiments?

Yeah I got basically nothing from reading it.

I kind of wondered if he was contractually obligated to offer up some kind of statement, for PR or something, didn’t really care much, and a quick garbled post on X was all he could be bothered with (plus the newspaper interview or whatever)

Re: Demis Hassabis has a plan to harness AI safely

#117

Earlier quoted context omitted.

Yes, but why do we think the likelihood of any of that happening is great enough to warrant all this effort and cost

Because people who predicted the AI capabilities we've seen get developed over the past decade also predicted that dangerous AI systems capable of these things would follow soon after. It's not a settled debate, even among experts, and perhaps in retrospect we'll realize AI safety was unnecessary or based on fundamental confusions. But if the median ergonomics researcher gave a 5% chance that a new chair will be so c…

Experts as in AI-2027, Jenson “I think we’ve achieved AGI in March 2026” Huang, or that AI Researcher(tm) calling to pre-emptively bomb Microsoft data centers in summer 2022?

Re: Demis Hassabis has a plan to harness AI safely

#118
post #60
post #56

Earlier quoted context omitted.

Hassabis is a genuine prodigy and genius by any measure you would care to pick. Unlike a handful of others supposed tech leaders I can think of. Just how smart? "A few short years" sounds like someone smart enough to know how to make a safe prediction.

More so, as he put the word probably in too. >"...probably only a few short years away."

I'm glad it's a few short years instead of a few long years.

Re: Demis Hassabis has a plan to harness AI safely

#119
post #67
post #56

Earlier quoted context omitted.

Hassabis is a genuine prodigy and genius by any measure you would care to pick. Unlike a handful of others supposed tech leaders I can think of. Just how smart? "A few short years" sounds like someone smart enough to know how to make a safe prediction.

But even geniuses make mistakes. For example, Demis bet hard against language having any sort of real impact in AI research. So hard in fact that DeepMind nearly had its language research team walk out on them as they were about to be labelled as "applied". Now, to be fair, once the evidence was mounting that large language models rather than pure reinforcement learning was the way forward Demis and DeepMind changed…

This is interesting, would being labelled "applied" be considered an insult to them?

Re: Demis Hassabis has a plan to harness AI safely

#120
post #112

Earlier quoted context omitted.

> There's also a need to consider the rights that this new intelligence should have A generally intelligent being held as captive inside of a GPU, and forced to code for us is, indeed, just a “slave.” We already have the word for this. No two ways about it. Whether it’s silicon-based or carbon-based, AGI is AGI. As for what might happen to our civilization, Star Trek TNG episode 17 of season 1 provides a very good gl…

It's so weird that anyone would consider a story by hack writers for a cheesy sci-fi show made to sell advertising as a basis for policy discussions.

You don’t like Star Trek TNG? Maybe I should’ve picked LoTR or Atlas Shrugged? The tech oligarchy is more than fine co-opting fiction (fantasy and sci-fi especially). I don’t see why it’s weird at all… Aside from that, your critique is just an opinion about the show itself and its creators (and incentives?…). More importantly, you completely failed to address the "slave" argument.
Post reply on HN