LLMs and their efficiency, can they really replace humans?
Evaluating the Energy-Efficiency of the Code Generated by LLMs
AI bros won't hype this up for the news for sure, but 480x energy doesn't sound optimistic enough for replacement.
LLMs don't have reasoning nor internal logic. If you take a look at the "thinking" feature AIs like Gemini introduced, this becomes even more obvious. In order to have the most basic type of analysis possible, it must hallucinate an entire context window to force the language model to reach a specific conclusion.
There's zero world in which LLMs replace humans. They might, temporarily, be convincing enough to trick a few CEOs... But that period of time won't last long.
Now, a human being assisted by AI on Microsoft Word or their Python IDE, sure.
extrapolate. what in 10 years?
If they're still LLMs? Nothing much changes.
as i've read somewhere, finite state machines cannot be sentient, or "intelligent" as we expect them to be. An LLM can not learn new things once trained. I'm waiting for a new breakthrough in this field, to be fully convinced about getting replaced.
https://arxiv.org/abs/2505.13763
new papers come out by the hour (literally) and i cant keep up. xD
or rather ask ai, it can give a better answer than me. xD