zlacker

[parent] [thread] 10 comments
1. jay-ba+(OP)[view] [source] 2024-05-18 01:32:04
When it comes to AI, as a rule, you should assume that whatever has been made public by a company like OpenAI is AT LEAST 6 months behind what they’ve accomplished internally. At least.

So yes, the insiders very likely know a thing or two that the rest of us don’t.

replies(4): >>vineya+C >>ein0p+L2 >>solida+Sh >>HarHar+dV
2. vineya+C[view] [source] 2024-05-18 01:42:41
>>jay-ba+(OP)
I understand this argument, but I can't help but feel we're all kidding ourselves assuming that their engineers are really living in the future.

The most obvious reason is costs - if it costs many millions to train foundation models, they don't have a ton of experiments sitting around on a shelf waiting to be used. They may only get 1 shot at the base-model training. Sure productization isn't instant, but no one is throwing out that investment or delaying it longer than necessary. I cannot fathom that you can train an LLM at like 1% size/tokens/parameters to experiment on hyper parameters, architecture, etc and have a strong idea on end-performance or marketability.

Additionally, I've been part of many product launches - both hyped up big-news-events and unheard of flops. Every time, I'd say that 25-50% of the product is built/polished in the mad rush between press event and launch day. For an ML Model, this might be different, but again see above point.

Sure products may be planned month/years out, but OpenAI didn't even know LLMs were going to be this big a deal in May 2022. They had GPT-2 and GPT-3 and thought they were fun toys at that time, and had an idea for a cool tech demo. I think that OpenAI (and Google, etc) are entirely living day-to-day with this tech like those of us on the outside.

replies(1): >>HarHar+mW
3. ein0p+L2[view] [source] 2024-05-18 02:12:48
>>jay-ba+(OP)
If they had anything close to AGI, they’d just have it improve itself. Externally this would manifest as layoffs.
replies(1): >>int_19+U6
◧◩
4. int_19+U6[view] [source] [discussion] 2024-05-18 03:22:46
>>ein0p+L2
This really doesn't follow. True AGI would be general, but it doesn't necessarily mean that it's smarter than people; especially the kind of people who work as top researchers for OpenAI.
replies(1): >>ein0p+rf
◧◩◪
5. ein0p+rf[view] [source] [discussion] 2024-05-18 06:07:35
>>int_19+U6
I don’t see why it wouldn’t be superhuman if there’s any intelligence at all. It already is superhuman at memory and paying attention, image recognition, languages, etc. Add cognition to that and humans basically become pets. Trouble is nobody has a foggiest clue on how to add cognition to any of this.
replies(1): >>int_19+9k
6. solida+Sh[view] [source] 2024-05-18 06:45:08
>>jay-ba+(OP)
But you also have to remember that the pursuit of AGI is a vital story behind things like fundraising, hiring, influencing politicians, being able to leave and raise large amounts of money for your next endeavor, etc.

If you've been working on AI, you've seen everything go up and to the right for a while - who really benefits from pointing out that a slowdown is occurring? Who is incentivized to talk about how the benefits from scaling are slowing down or the publicly available internet-scale corpuses are running out? Not anyone who trains models and needs compute, I can tell you that much. And not anyone who has a financial interest in these companies either.

◧◩◪◨
7. int_19+9k[view] [source] [discussion] 2024-05-18 07:23:46
>>ein0p+rf
It is definitely not superhuman or even above average when it comes to creative problem solving, which is the relevant thing here. This is seemingly something that scales with model size, but if so, any gains here are going to be gradual, not sudden.
replies(1): >>ein0p+Mn
◧◩◪◨⬒
8. ein0p+Mn[view] [source] [discussion] 2024-05-18 08:19:09
>>int_19+9k
I’m actually not so sure they will be gradual. It’ll be like with LLMs themselves where we went from shit to gold in the span of a month when GPT 3.5 came out.
replies(1): >>int_19+nM1
9. HarHar+dV[view] [source] 2024-05-18 14:32:16
>>jay-ba+(OP)
Sure, they know what they are about to release next, and what they plan to work on after that, but they are not clairvoyants and don't know how their plans are going to pan out.

What we're going to see over next year seems mostly pretty obvious - a lot of productization (tool use, history, etc), and a lot of efforts with multimodality, synthetic data, and post-training to add knowledge, reduce brittleness, and increase benchmark scores. None of which will do much to advance core intelligence.

The major short-term unknown seems to be how these companies will be attempting to improve planning/reasoning, and how successful that will be. OpenAI's Schulman just talked about post-training RL over longer (multi-reasoning steps) time horizons, and another approach is external tree-of-thoughts type scaffolding. These both seem more about maximizing what you can get out of the base model rather than fundamentally extending it's capabilities.

◧◩
10. HarHar+mW[view] [source] [discussion] 2024-05-18 14:43:02
>>vineya+C
> I think that OpenAI (and Google, etc) are entirely living day-to-day with this tech like those of us on the outside.

I agree, and they are also living in a group-think bubble of AI/AGI hype. I don't think you'd be too welcome at OpenAI as a developer if you didn't believe they are on the path to AGI.

◧◩◪◨⬒⬓
11. int_19+nM1[view] [source] [discussion] 2024-05-18 22:34:54
>>ein0p+Mn
Much of what GPT 3.5 could do was already there with GPT 3. The biggest change was actually the public awareness.
[go to top]