Is the AI Hype Over Yet?
Has the hype bubble for AI finally popped?
If the recent worries and drop in NVIDIA’s share price are anything to go by as an indicator, then you would certainly think so. Zoom out a little though and despite the recent 30sih percent drop (most of which has recovered) NVIDIA is still worth multiples of what it was only a year ago. So maybe not.
Is the AI hype train slowing down at least then? I feel like we are certainly seeing signs of this with increasing news and social media content focusing on the limitations of LLMs rather than spinning the OpenAI lead narrative of an impending AI apocalypse, oh sorry I mean singularity.
The sooner we see an implosion, or at the very least a correction, in the AI bubble the better off we will be in the long term. LLMs are a great bit of tech. The biggest problem with them isn’t even the hallucinations they produce. No, the biggest problem with them is the hallucinations that their creators seem to be suffering from.
If AI companies could stop pretending that we’re 3 months away from living like “The Jetsons” and focus on figuring out how on earth LLMs can actually be made cost effective, we might even produce something financially viable rather than just choking the planet to create more cat pictures.
Instead of claiming to replace software developers for a minute, instead and give me better integration of a Copilot like tool that looks at the compiler output, test results or perhaps even log output on the developer machine. That would be much appreciated. Please and thank you. Increase its context window to include the whole project, maybe even the third-party libraries and make it reuse rather than rewrite code. Let me train models with proprietary and confidential codebases and run models locally so they can be used on such codebases.
We don’t seem to even be close to a coding assistant that can do the above but apparently software developers will be out of a job in less than 5 years. Pull the other one mate, its got bells on.
