YHK - yet another hacker news clone

> But traditional technologies lack cognitive processes: they do not perceive, evaluate, or solve problems

Neither do LLMs, which are "billions of dice" -word-guessers, still to this day.

To me, it feels like the opposite: a useful tool that's been wrongly marketed as something more. LLMs are very helpful in many different contexts. Sure, you could also use them to discuss your intimate issues or make your decisions for you, but you could also do that with Eliza back in the day if you wanted to. Wasn't a good idea then, isn't a good idea now.

Hmm.. agent.as_tool() looks like a tool for me

I try not to put much stock in black-and-white opinions because I think the answer is rarely that simple.

Humans are probably the least tool liking in all known universe. Yet, humans were used like tools everywhere every time in the society.

AI is a tool for sure.

Yeah, this "we have to adapt like we always did" feels very thought-terminating. I mean, you don't adapt to a nuclear bomb falling over your city.

sigh No, it doesn't have consciousness, nor have human thoughts, etc. We (humans) are just _really_ good at anthromorphizing and pattern-matching. We do it to our animals, assigning human feelings to their expressions, and so of course we'll do it to the tool that outputs in our language.

> But beyond that, AI can change, improve itself, and even alter part of its nature. Tools don't do that.

I mean what do you call it when a package adds new functionality in v1.1 vs 1.0?

Improving itself is hard to see. I don’t as a developer using opus 5.5 see it improving itself. It’s great, and I expect the next opus to be better, but while I have this version it seems pretty much the same.

The Internet is also just a series of tubes... but there's emergent properties that come out of any sufficiently complex system.

I think the framing provided in this creative writing is quite disagreeable. But I can understand why it feels that way. The important thing is to comprehend that it's not actually that way. The Internet is just a series of tubes... even though sometimes I refer to the Internet as an entity that can do some really funny stuff...

AI doesn't exist. LLMs do, and those are tools.

AI is a tool the same way humans are tools for other humans.

A corporation is basically human tools organized in hierarchical layers, each layer is human tools for the layer above to carry some tasks.

The "AI is a tool/is not a tool" pieces all rest very much on their definition of the word "tool".

This one rests on the definition of "herramienta". I don't have nearly good enough Spanish to understand if there are additional subtleties relating to how that word is used in that language.

(That said, it's pretty great that we can have a link to a Spanish thread from an English language forum and the language barrier is almost invisible.)

This would be true when software factories succeed. For now, AI needs initial input from a human

An expert system is a tool, and we've had expert systems in computing for ages now.

One could, perhaps, make the case that the inherent randomness in the LLM algorithm makes it a fundamentally different thing than an expert system.

... but we've had Monte Carlo analysis as a tool for ages too.

I'm not sure we promote the thing to "not a tool" by gluing a Monte Carlo algorithm onto an expert system.

To take a big step back from the LLM question and look at a similar space that perhaps sheds light on the question...

A lot of chip design these days involves a step where the desired circuit is described in a goals-and-constraints-oriented language and then fed to a layout algoirthm that is semi-deterministic and attempts to solve the optimization function based on specified goals (like fabrication cost, size, or power consumption).

The output of this process is a chip you can fab. The process is random (it's an important industry-standard step to capture the input seed, because if the chip ends up in something audited and the auditing firm needs to know how the chip was made, it's not good enough to shrug and say "Enh, we rolled some dice," they need the layout schema deterministically reproducible too).

I don't, generally, think anyone considers the fact we use these in the hardware space to be anything but using yet another tool. It's a tool that searches a solutions domain much faster than a human could doing something not too dissimilar from what a human would do (with perhaps a little human intuition thrown on the top to avoid obviously-pessimistic branches of exploration, but you can bake those shortcuts into the tool also).

Why does its capacity for autonomous action mean it isn't a tool, rather than that 'tool' is an insufficient description of its capacities and risks?

"AI is just a tool" seems like a thought terminating cliche.