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

AI 2027

ai-2027.com

Post reply on HN
Print view View original

211–220 of 641 posts

Re: AI 2027

#211

Post by Tossrock »

turnsout wrote: post #153
IshKebab wrote: post #31

This is hilariously over-optimistic on the timescales. Like on this timeline we'll have a Mars colony in 10 years, immortality drugs in 15 and Half Life 3 in 20.

IMO they haven't even predicted mid-2025. > Coding AIs increasingly look like autonomous agents rather than mere assistants: taking instructions via Slack or Teams and making substantial code changes on their own, sometimes saving hours or even days. Yeah, we are so not there yet.

That is literally the pitch line for Devin. I recently spoke to the CTO of a small healthtech startup and he was very pro-Devin for small fixes and PRs, and thought he was getting his money worth. Claude Code is a little clunkier but gives better results, and it wouldn't take much effort to hook it up to a Slack interface.

Re: AI 2027

#212

Post by Aurornis »

mchusma wrote: post #58
IshKebab wrote: post #31

This is hilariously over-optimistic on the timescales. Like on this timeline we'll have a Mars colony in 10 years, immortality drugs in 15 and Half Life 3 in 20.

I like that the "slowdown" scenario has by 2030 we have a robot economy, cure for aging, brain uploading, and are working on a Dyson Sphere.

The story is very clearly modeled to follow the exponential curve they show.

Like the drew the curve out into the shape they wanted, put some milestones on it, and then went to work imagining what would happen if it continued with a heavy dose of X-risk doomerism to keep it spicy.

It conveniently ignores all of the physical constraints around things like manufacturing GPUs and scaling training networks.

Re: AI 2027

#213

Post by Tossrock »

AlexandrB wrote: post #98
barbarr wrote: post #82

Earlier quoted context omitted.

It also ignores the possibility of plateau... maybe there's a maximum amount of intelligence that matter can support, and it doesn't scale up with copies or speed.

Or scales sub-linearly with hardware. When you're in the rising portion of an S-curve[1] you can't tell how much longer it will go on before plateauing. A lot of this resembles post-war futurism that assumed we would all be flying around in spaceships and personal flying cars within a decade. Unfortunately the rapid pace of transportation innovation slowed due to physical and cost constraints and we've made little pr…

The fact that it scales sub linearly with hardware is well known and in fact foundational to the scaling laws on which modern LLMs are built, ie performance scales remarkably closely to log(compute+data), over many orders of magnitude.

Re: AI 2027

#214

Post by dimitri-vs »

torginus wrote: post #143

Much has been made in its article about autonomous agents ability to do research via browsing the web - the web is 90% garbage by weight (including articles on certain specialist topics). And it shows. When I used GPT's deep research to research the topic, it generated a shallow and largely incorrect summary of the issue, owning mostly to its inability to find quality material, instead it ended up going for places li…

Interesting, I've hard the exact opposite experience. For example I was curious why in metal casting the top box is called the cope and the bottom is called the drag. And it found very niche information and quotes from page 100 in a PDF on some random government website. The whole report was extremely detailed and verifiable if I followed its links.

That said I suspect (and am already starting to see) the increased use of anti-bot protection to combat browser use agents.

Re: AI 2027

#215

Post by wkat4242 »

pydry wrote: post #182
KaiserPro wrote: post #151

> AI has started to take jobs, but has also created new ones. Yeah nah, theres a key thing missing here, the number of jobs created needs to be more than the ones it's destroyed, and they need to be better paying and happen in time. History says that actually when this happens, an entire generation is yeeted on to the streets (see powered looms, Jacquard machine, steam powered machine tools) All of that cheap labour…

>History says that actually when this happens, an entire generation is yeeted on to the streets History hasnt had to contend with a birth rate of 0.7-1.6. It's kind of interesting that the elite capitalist media (economist, bloomberg, forbes, etc) is projecting a future crisis of both not enough workers and not enough jobs simultaneously.

I don't really get the American preoccupation with birth rates. We're already way overpopulated for our planet and this is showing in environmental issues, housing cost, overcrowded cities etc.

It's totally a great thing if we start plateauing our population and even reduce it a bit. And no we're not going extinct. It'll just cause some temporary issues like an ageing population that has to be cared for but those issues are much more readily fixable than environmental destruction.

Re: AI 2027

#216

Post by torlok »

KaiserPro wrote: post #151

> AI has started to take jobs, but has also created new ones. Yeah nah, theres a key thing missing here, the number of jobs created needs to be more than the ones it's destroyed, and they need to be better paying and happen in time. History says that actually when this happens, an entire generation is yeeted on to the streets (see powered looms, Jacquard machine, steam powered machine tools) All of that cheap labour…

Hayek has been lobbied by US corporations so hard for so long that regular people treat the invisible hand of the market like it's gospel.

Re: AI 2027

#217

Post by ffsm8 »

neural_thing wrote: post #121
ffsm8 wrote: post #90

Earlier quoted context omitted.

People have been predicting the singularity to occur sometimes around 2030 and 2045 waaaay further back then 2015. And not just by enthusiasts, I dimly remember an interview with Richard Darkins from back in the day... Though that doesn't mean that the current version of language models will ever achieve AGI, and I sincerely doubt they will. They'll likely be a component in the AI, but likely not the thing that "driv…

Vernor Vinge as much as anyone can be credited with the concept of the singularity. In his 1993 essay on it, he said he'd be surprised if it happened before 2005 or after 2030 https://edoras.sdsu.edu/~vinge/misc/singularity.html

Fwiw, that prediction was during Moore's law though. If that held until now, CPUs would run laps around what our current gpus do for LLMs.

Re: AI 2027

#218

Post by wkat4242 »

danpalmer wrote: post #185
IshKebab wrote: post #31

This is hilariously over-optimistic on the timescales. Like on this timeline we'll have a Mars colony in 10 years, immortality drugs in 15 and Half Life 3 in 20.

These timelines always assume that things progress as quickly as they can be conceived of, likely because these timelines come from "Ideas Guys" whose involvement typically ends at that point. Orbital mechanics begs to disagree about a Mars colony in 10 years. Drug discovery has many steps that take time, even just the trials will take 5 years, let alone actually finding the drugs.

Didn't the covid significantly reduce trial times? I thought that was such a success that they continued on the same foot.

Re: AI 2027

#219

Post by ivraatiems »

Though I think it is probably mostly science-fiction, this is one of the more chillingly thorough descriptions of potential AGI takeoff scenarios that I've seen. I think part of the problem is that the world you get if you go with the "Slowdown"/somewhat more aligned world is still pretty rough for humans: What's the point of our existence if we have no way to meaningfully contribute to our own world?

I hope we're wrong about a lot of this, and AGI turns out to either be impossible, or much less useful than we think it will be. I hope we end up in a world where humans' value increases, instead of decreasing. At a minimum, if AGI is possible, I hope we can imbue it with ethics that allow it to make decisions that value other sentient life.

Do I think this will actually happen in two years, let alone five or ten or fifty? Not really. I think it is wildly optimistic to assume we can get there from here - where "here" is LLM technology, mostly. But five years ago, I thought the idea of LLMs themselves working as well as they do at speaking conversational English was essentially fiction - so really, anything is possible, or at least worth considering.

"May you live in interesting times" is a curse for a reason.

Re: AI 2027

#220

Post by wkat4242 »

noncoml wrote: post #40

2015: We will have FSD(full autonomy) by 2017

Well, Teslas do have "Full Self Driving". It's not actually fully self driving and that doesn't even seem to be on the horizon but it doesn't appear to be stopping Tesla supporters.
Post reply on HN