Live data from Hacker News

The Gentle Singularity

blog.samaltman.com

491–500 of 557 posts

Re: The Gentle Singularity

#492
post #485

> Scientific progress is the biggest driver of overall progress > There will be very hard parts like whole classes of jobs going away, but on the other hand the world will be getting so much richer so quickly that we’ll be able to seriously entertain new policy ideas we never could before Real wages haven’t risen since 1980. Wealth inequality has. Most people have much less political power than they used to as wealth…

> Real wages haven’t risen since 1980. Do people really believe that? I think either people have too rosy view of 80s or consider that real wages should also adjust for lifestyle inflation.

Yes. It’s even a part of Ray Dalio’s speeches on the topic. Here is one example where he mentions it: https://www.linkedin.com/pulse/why-how-capitalism-needs-refo...

Re: The Gentle Singularity

#493
post #343
post #147

Earlier quoted context omitted.

And self driving cars. Real time ray tracing. WiFi. Cloud computing. starlink. Crispr. mRNA vaccines. Blockchain. Voice assistants.

All the significant tech on that list is still "two years away from being two years away".

What do you mean? I take a self driving car around every week. I got a mRNA vaccine that enabled us to get out of a pandemic much quicker. You can get fast internet in the middle of nowhere via starlink.

Re: The Gentle Singularity

#494

Earlier quoted context omitted.

I hate to enter this discussion, but learning based on a small number of examples is called few-shot learning, and is something that GPT-3 could already do. It was considered a major breakthrough at the time. The fact that we call gradient descent "learning" doesn't mean that what happens with a well-placed prompt is not "learning" in the colloquial sense. Try it: you can teach today's frontier reasoner models to do…

Are you talking about teaching in the context window or fine tuning? If it is the context window, then you are limited to the size of said window and everything is lost on the next run. Learning is memory, what you are describing is an llm being the main character in the movie Momento, I.e. no longterm memories past what was trained in the last training run.

There's really no defensible way to call one "learning" and the other not. You can carry a half-full context window (aka prompt) with you at all times. Maybe you can't learn many things at once this way (though you might be surprised what knowledge can be densely stored in 1m tokens), but it definitely fits the GP's definition of (1) real-time and (2) based on a few examples.

Re: The Gentle Singularity

#495

> Scientific progress is the biggest driver of overall progress > There will be very hard parts like whole classes of jobs going away, but on the other hand the world will be getting so much richer so quickly that we’ll be able to seriously entertain new policy ideas we never could before Real wages haven’t risen since 1980. Wealth inequality has. Most people have much less political power than they used to as wealth…

> The utopia Altman describes won’t appear.

Sure it will, as far as Altman is concerned. To make the whole post make sense, add "... for the rich" where appropriate.

Re: The Gentle Singularity

#496

> Already we live with incredible digital intelligence, and after some initial shock, most of us are pretty used to it. Very quickly we go from being amazed that AI can generate a beautifully-written paragraph to wondering when it can generate a beautifully-written novel; It was probably around 7 years ago when I first got interested in machine learning. Back then I followed a crude YouTube tutorial which consisted o…

What's your take on Anthropic's 'Tracing the thoughts of a large language model'? [0] > To write the second line, the model had to satisfy two constraints at the same time: the need to rhyme (with "grab it"), and the need to make sense (why did he grab the carrot?). Our guess was that Claude was writing word-by-word without much forethought until the end of the line, where it would make sure to pick a word that rhyme…

Generalize the concept from next token prediction to coming tokens prediction and the rest still applies. LLMs are still incredibly poor at symbolic thought and following multi-step algorithms, and I as a non-ML person don't really see what in the LLM mechanism would provide such power. Or maybe we're still just another 1000x scale off and symbolic thought will emerge at some point.

Me personally, I expect to see LLMs to be a mere part of whatever will be invented later.

Re: The Gentle Singularity

#497

Earlier quoted context omitted.

Nuclear plants have other long tails costs, decommissioning and waste containment.

So do coal plants. Interestingly approximately nobody seems to be worried about the thousands of coal ash ponds leaching radioactive material, toxic heavy metals and teratogenic compounds into ground water on a regular basis - while storing spent fuel element in a salt mine thousands of miles away engenders outrage…

I have never worried about a coal plant in Ukraine. I lived at a place that was potentially downwind of Chernobyl and inside the projected bad-things-happening plume. The scale of a nuclear disaster well exceeds the scale of a coal plant disaster.

Many things are cheap when you ignore externalities.

Re: The Gentle Singularity

#498

Earlier quoted context omitted.

Citation needed, my understanding was that housing prices are being driven up by real estate owners/agencies buying tons of property to either rent at extortionary prices or sit on until they sell for a higher price to the next sucker. Also stuff like the RealPage scandal where they simply illegally collude: https://news.ycombinator.com/item?id=41330007 I think the idea of a law that only allows a limited number of o…

Housing prices are being driven up by the fact that we don't build enough housing. That should be the end of the discussion. What you describe is just a symptom of not having enough housing. If we just built more housing then there wouldn't be any business angle for the corporations buying up housing.

Well, why dont we build more housing?

Re: The Gentle Singularity

#499

Earlier quoted context omitted.

You at least had the option of buying one that did not run on dead dinosaurs.

Dead pedestrians are not a suitable replacement.

Ooh, a self-driving-into-obstacles car that runs on dead pedestrians. With garbage truck style arms to lift the bodies automatically.

Re: The Gentle Singularity

#500

Earlier quoted context omitted.

No, they haven't , at least nowhere near the scope of the hype. The promise in the early 2010s was that in 10-20 years, human driving would be obsolete, that professional drivers would be out of a job, and that all land-based transportation would be overhauled. None of this is even close to happening. Waymo is impressive tech, for sure, but what they're proving is that this is just not currently solvable as a general…

The optimists got the timeline wrong but they were closer to the truth than those who said it would never happen. I have no idea how close AGI is but there is clearly no AI winter around the corner as some HNers have been predicting[0]. [0] https://news.ycombinator.com/item?id=23886325

Solving a small subset of the problem != solving the problem.

I'll believe self-driving is solved when I can order an automated ride to a BLM/forest service dirt road I only have GPS coordinates for and that has cell service on the hilltop only.

Post reply on HN