I'm going out onto a limb here. Calling Artificial Intelligence "pee-pee" is, probably, going to generate much hilarity. I myself cannot stop smiling about it.
Very useful information! Thank you. Personally, I use LLM as a nextgen search engine and grammar checks. I'm from the south and speak in past participles.
Catching up on your old articles. Just want to add:
> "These labels are just an attempt to describe a set of processes, that are really unlike anything that a human does. The problem with language is that we immediately begin to anthropomorphise it in order to make sense of it. And in the process of doing that, it is the audience or the listener that breathes life into the Frankenstein monster."
This bit really hits the nail on the head. It's people's idea ("belief") about AGI leading them into a psychopathic feedback loop, and this is intentionally cultivated in the way companies market it. The people create their own dragon by fearing the dragon, and they lead society down the wrong path by "reacting" to a scarecrow that doesn't exist. I hate to use the term "psyop," but I mean, damn. If it's not one, it sure is convenient for some people.
"We have nothing to fear but fear itself." I really don't like FDR, but I appreciate the quote more as I get older.
I've always thought "AI" should really stand for Artificial Imitator more than anything else. I'm with you on the fact that the more you peel back the layers, the less impressive the "intelligence" appears and the more you see the weaknesses. A good way to temper the discussion and move away from the hype and abstract fear is to demystify it by going through what you just did. But therein lies the problem IMO--who has time for ML-101 so to speak, and the requisite CS? Even just the manual labor data cleanup and massive crowdsourced data labeling is well hidden from the public (like OpenAI paying 10k Kenyans to do that job at $2/hr) Heck even now no one in the field wants to do all that and more like you from scratch, and would rather use a pre-trained/turnkey model that would maybe just use some fine tuning
I'm going out onto a limb here. Calling Artificial Intelligence "pee-pee" is, probably, going to generate much hilarity. I myself cannot stop smiling about it.
Very useful information! Thank you. Personally, I use LLM as a nextgen search engine and grammar checks. I'm from the south and speak in past participles.
Catching up on your old articles. Just want to add:
> "These labels are just an attempt to describe a set of processes, that are really unlike anything that a human does. The problem with language is that we immediately begin to anthropomorphise it in order to make sense of it. And in the process of doing that, it is the audience or the listener that breathes life into the Frankenstein monster."
This bit really hits the nail on the head. It's people's idea ("belief") about AGI leading them into a psychopathic feedback loop, and this is intentionally cultivated in the way companies market it. The people create their own dragon by fearing the dragon, and they lead society down the wrong path by "reacting" to a scarecrow that doesn't exist. I hate to use the term "psyop," but I mean, damn. If it's not one, it sure is convenient for some people.
"We have nothing to fear but fear itself." I really don't like FDR, but I appreciate the quote more as I get older.
Would be interested in your thoughts on Consciousness and Computational Theory of Mind.
I've always thought "AI" should really stand for Artificial Imitator more than anything else. I'm with you on the fact that the more you peel back the layers, the less impressive the "intelligence" appears and the more you see the weaknesses. A good way to temper the discussion and move away from the hype and abstract fear is to demystify it by going through what you just did. But therein lies the problem IMO--who has time for ML-101 so to speak, and the requisite CS? Even just the manual labor data cleanup and massive crowdsourced data labeling is well hidden from the public (like OpenAI paying 10k Kenyans to do that job at $2/hr) Heck even now no one in the field wants to do all that and more like you from scratch, and would rather use a pre-trained/turnkey model that would maybe just use some fine tuning
💬 I had essays half written [...], but I’m not sure any of that really matters or is pertinent.
The only way to find out is to fit on the missing halves, and onwards onto launchpad 😊 What do you think LLMs cum PPs are for‽