[dagelf]

Another know-it-all calling AI a bubble. Lets revisit this, not in 10 years, but in 5. Things the article misses: "Bubbles" happen for more reasons than hype. There are moats, just not in the obvious or same old places. Funny thing is, an LLM can probably write a better TL;DR, that captures it better, not to mention a more insightful article.

[skerit]

This article seems to completely misread what the DeepSeek release actually represents. The author uses it as evidence of the AI bubble bursting, but DeepSeek demonstrates exactly the opposite: AI becoming more efficient and practical, not less valuable.

A small company just released a model that matches leading proprietary systems at 1/30th the running costs and fraction of the training costs ($6M vs $60M+). Yes, this challenges the "moat" of big tech companies, but that's not a bubble bursting - it's a technology becoming more accessible and practical. The author conflates "big tech losing its monopoly" with "AI being a fad."

[CharlesW]

Agreed. I consider myself an AI skeptic, but this reads like the first-draft ramblings of someone who's never heard of the Gartner Hype Cycle.

[hooli_gan]

https://en.m.wikipedia.org/wiki/Gartner_hype_cycle

[mbrumlow]

Yah. It might write a better TL;DR, but it will always be a tool engineers use.

AI is a bubble if you consider the end goal of AI to be a 1:1 replacement for humans.

AI is not a bubble if consider the more likely end goal of AI being just another tool. A really good tool but just a tool real indulgence will use. We have seen this many times in the past with tech. While some people did loose jobs over time it evened out and eventually made more jobs. Just look at how many jobs today exist simply because computers exist, a hardware device that was to then replace the modern worker.

[cheevly]

It will be as much a ‘tool’ as human intelligence is a tool.