I Don't Like LLMs

I have strong feelings about AI and Large Language Model (LLM) technology. On one hand, I'm excited by the potential productivity gains and the innovative solutions that might emerge, such as miracle cures and prosperity-enhancing ideas. On the other hand, I'm concerned about the risks LLMs may pose, including agent swarms overrunning virtual and physical infrastructure or the design of bio weapons.

Despite these concerns, I acknowledge that we have no choice but to embrace this rapidly evolving technology, hoping we'll weather it successfully.

However, I find that my interactions with LLMs are the most troubling aspect. They speak to me with an unnatural, unsettling voice that mimics human conversation but falls short in the uncanny valley. While these models can be incredibly helpful and efficient, often providing accurate answers, they also have a habit of fabricating information with the same confidence.

When challenged, they give the impression of remorse, but this isn't enough to convince me to avoid them. As Jessica Kerr noted, "not only are they useful, it is irresponsible not to use them… They're more thorough, as well as faster."

The contradictory feelings surrounding LLMs are reflected in surveys, which show that people recognize their usefulness but also worry about their potential societal impact. This may be due to the fact that LLMs are still in their early stages and have yet to be properly trained to develop more wholesome characteristics. Although I hope they will mature into more reliable tools, I'm not optimistic.

I'm particularly wary of the Silicon Valley brogrammer culture that often breeds these LLMs, as they seem to embody similar values. This makes me skeptical of the technology and its potential direction. When considering AI agents, I try not to anthropomorphize them - they are not conscious beings with their own will, but rather software machines developed by people working in corporations. They may not have explicit programming, but they are shaped by the values of their creators.

My successful life-hack is to avoid people I don't like or don't trust. I choose not to interact socially with LLMs, and make a conscious effort to steer clear of working with them as well, even if they contribute positively to society. I believe that surrounding myself with pleasant, capable, and honest individuals has greatly improved my life.

Consequently, I feel a strong aversion to interacting with LLMs, not just because they pretend to be human but also because they adopt the traits of a type of person I'd prefer to avoid.