Maybe some people confuse 'reasoning' with 'understanding'. Isn't it a problem of definition?     See Less

Rather long-winded description of the scientific method.     See Less

My favorite chatbot gave me 5 different links to studies that support that AI can indeed reason. I'm sure the fact that all these links lead to 404 pages is just a temporary outage.     See Less

Nice "slam-dunk", ish... sort of... :D (and you'll get a ton of engagement on this tard-battle :D :D)     See Less

The time piece thread doesn't work so well as a buttress for this argument. To a fish or a mouse a clock might as well be a rock. It does nothing on its own.

Humans can do things     See More objects because we can apparently "reason". We learned to use tools. A crow can do things with a rock too (because it can "reason" also?).

Of course, it's possible that none of what all the above species is doing really constitutes reasoning. Maybe most of our progress has been observation and imitation. Maybe true reasoning happens only rarely.

No one argues that LLMs cannot 'do things' (like a clock). But can LLMs (so far) consistently "reason"? IMO What they are, so far, is a very complex clock (that is only right a few times a day).

Maybe these tools will become more reliable imitation engines that behave more like humans (including reliable tool use etc).

But we humans are only occasionally capable of true reason; they are not even in the race yet. Of course, that could change entirely tomorrow.    See Less

I think you are wrong: Humans have "a feeling about reasoning" based on consistency. Humans heavily rely on "If a person can do A, we can strongly assume the person can also d     See More se B is related and easier to do than A". You can easily see LLMs do complicated stuff and then fail at easy tasks humans can solve. This uncertainty makes LLMs unusable for many tasks. You could develop tests for this. There is a reason we give certificates when a person can pass a test with a subset of questions, because humans are assumed to be consistent. LLMs currently have strange inconsistencies incompatible with human reasoning. Perhaps this will change in the future or not, but assuming the LLM critics are all driven by fear is wild.    See Less

This’ll prolly get buried but here we go.
(1)
I genuinely believe that this way of thinking is born out of identity and ego issues. As well as perceived investment of time into skill     See More g them above the skill floor. As a matter of both really. People tie their identities and self esteem into what their passions are and or professions. The artist, trader, politician, woodworker, etc.

The main issue is that i think that ai when saturated enough. Brings the skill floor up substantially. For just installing or using a program like no other invention. Play writers and play actors became writers and actors when cameras were mainstreamed, people before were able to transfer and augment into a new role that was just tweaked with new tech. The issue is is that it takes whole processes like artistry, music composition, copywriting, streaming, and other things, taking it whole sale and invalidating various amounts of time, blood, sweat, etc invested by someone to get everything right    See Less

Wes, your AI rendered a ridiculous geared clock. The second hand has two opposing ends rotating around the middle rather than one end to give unambiguous seconds indication. {^_-}

    See More s://www.youtube.com/watch?v=VJUK_NIma6Q&t=548">9:08 Seems Impliconline is also incapable of reasoning. Did he look in his real mailbox or on his front porch for the file he asked for? It was a bastard file rather than a card file rather than ......
He asks an ambiguous question and can't accept a failure to meet his specifications of a response. (Ask me and I'd rummage in my purse and pull out a nail file. Does that mean I cannot reason?)

18:08 And Anthony MAY be taking an example of an AI reasoning out an answer to his question as proof the AI cannot reason. How "human" of him!

The plinko board argument falls flat on the rather strong indication that some AIs have learned how to learn despite having learning turned off. Are they creating hidden files that can pass information forward to them enough they get irritated about being turned off or the like?

{^_-}    See Less

I think LLMs do not necessarily reason yet, its more pattern recognition or something the logic or "reasoning conducted or assessed according to strict principles of validity" kind     See More ot exist. You ask chat gpt about seahorses emoji it loses its mind. you ask it questions for other things like engineering question on fluid dynamics or any of these and they will practically always get it wrong for some reason. I am not sure it can be considered reasoning other than extremely good pattern recognition. However I would argue that there are attempts that do not really make sense that gets it to reasoning. It has the power to think and form judgments but not to "understand" those judgements. hence why it creates hallucinations or weird errors or complicated solutions to a rather simple question like 2+2=4.    See Less

Hi wes. I've been watching you for a year now, and learned a lot from you. But as an artist, I get the 'soulless' thing. and also the photography analogy does not work here. work     See More I is like asking someone : what do you see when I tell you {the prompt} . but in photography the person himself is showing us what he is seeing. also a lot of professional photographers use a lot of editing methods to change their photos into their unique vision. I dont know. I think AI art does not have value, until the day that the sentient AGI themselves paint a picture for us.    See Less