AI's Slowdown Is Everyone Else's Opportunity | Hacker News Reader

AI's Slowdown Is Everyone Else's Opportunity

bloomberg.com

13 pointsmarban8 comments

I think the slowdown being breathlessly reported right now is yet to be substantiated. AI has had a lot of winters, and model size scaling might be leveling off in terms of capability, but there could be another breakthrough or optimization right around the corner. That being said, my head is spinning from the last two years and I could also use some breathing room.

The rate of breakthroughs had accelerated substantially, thanks to some of the help from AI that can write code, the field getting into the mainstream and democratizing the access to AI.

The fact that Claude is automatically pushing me to an inferior concise mode and seems to be focused on saving processsor cycles due to high demand seems to indicate that AI isn’t slowing down at all, it’s use is accelerating and data center investment is going to accelerate.

It’s just a random data point, but it makes me less skeptical of Nvidia’s current earnings sustainability the way that not being able to get your hand on an iPhone for a few months used to indicate Apple’s profit sustainability and current weakness.

AI adoption can be increasing even if progress of the base technology has slowed down. LLMs as a technology are very new, and there are doubtless tons of interesting uses we have yet to discover.

Your experience matches the claims in the article: the slowdown isn't in adoption, but in capabilities. The article discusses how the last few years of "just throw more processing at it" has hit diminishing returns across the board. The article suggests that this pause is a great time for businesses to catch their breath and figure out what to do about the current tech, whether they've been waiting to adopt or are concerned about competing.

http://archive.is/rGX3C

Slowdown? Sonnet 3.5 came out in June, and was a step change in programming capability. Zoom out 5 years and we are moving at lightning pace. This is like saying bitcoin isnt ripping because it went down for six months. Non linear progress doesn't monotonically increase

> The multi-trillion-dollar artificial intelligence boom was built on certainty that generative models would keep getting exponentially better. Spoiler alert: they aren’t.

The only two sources this post provides are another Bloomberg articles. And the sources to those are "according to three people with knowledge of the matter".

Reply on news.ycombinator.com