Notes for 1 day challenge article Subject: my view on the value of AI, its AI products, and the future viablity of it AI in general --> not necessarily LLMs - challenge started 1400 Friday + notes section finished ~1612 --> that's making this messy file + started writing 1730-2008 --> mental fatigue kicked in +> finished "Information veracity and reliability" - started writing again at 1200 the next day - published 22:16 - comparison to .COM bubble - in fad nature + everything that said .COM whether it needed to or not stocks go up, everything that did not, stock go down. Because the market is only as rational as the people within it. I see the same thing now, every company has some kind of AI bullshit service. A useless chat bot, or a tool that just redresses ChatGPT. Duck.AI. - fallout of "Disruptive Technologies" --> the system pays the cost, and the cost trickle down to the public AI hallucinating case law references https://www.youtube.com/watch?v=bGn1SZuhTys Problem w/ Hallucinations --> can not rely on anything that AI actually tells you. +> so what use is it? Same as wikipedia. a stepping off point to get you into a subject. +> both wikipedia and ChatGPT cite their sources, and it's in those sources that more accurate information can be found. - reference Wikipedia.org in terms that it can not be trusted due to its Vox Populi nature & AI is aggregated Vox Populi - e.g. inaccurate information repeatedly posted multiple times will likely rank higher in AIs algorythms then correct information listed sparingly. + also the AI pulls from poor quality sources such as Reddit. +> where information is largely filtered through a political/tribal lens. Where more factional information will be downvoted and de-ranked for information that is deemed preferable. TOY example: Transwomen are women, Transwomen are men. One is factual, the other is feel good. Depending on the subbreddit that you are on. One or the other will be the officially accepted answer. In addition reddit due to their moderators and supermods politics one answer is significantly more likely than the other. Same with Wikipedia. AI in education --> notable negative effects on children who use it to cheat themselves out of learning. +> soultion to the AI problem: more AI --> AI to test for AI use in submitted course work. +> my solution: BACK-TO_BASICS. bring all you referenhttps://www.youtube.com/shorts/KzjHNSRgHJ8ce materials into class hand write your essays in a designated long session. Children seem to be less able to read and write, much less think critically --> Ipad upbringing generation, --> lazy parenting +more technology to constantly distract, as well as do cognitive labour. For a child who lacks awareness and guidance, it is all too easy to slide down the path of least resistance, and like rainwater you will end up the gutter should you follow it for long enough, or lost at sea. Bewildered by a world you are unprepared for. My D&D images: - AI making poor quality images --> finger inconsistency - AI not following prompts --> an inability to count --> too many train cars - can not put a sword in a characters hand without being babied with example image mockup + otherwise wastes too many prompts (tokens?) Then there is what AI does to the Internet - Search engines now have AI summaries for search queries. + this means that the use does not need to actually visit the websites to access their information +> this takes away 3rd party websites profit incentive --> i.e they make no money ++> lowing the available site quantity and quality +> which AI uses as its source --> like a snake eating its tail, or a farmer that poisons his own land to grow crops ==> it will lead to poorer future harvests - socially: social media bot accounts. Used for facilitating scams. and for pushing political narratives by interested parties. It makes social media considerably less trustworthy, and worth while. - image generation does a similar thing to artists. Harvests and uses their work to created its images. While subverting their market demand. >> I use AI to generate NPCs for my D&D campaign does that make me a hypocrit? No, listing out what I think is happeneing is not the same as saying that I care. In addition I follow that piracy logic of pirating media =/= does not equal a loss of a sale. The pirate was never going to buy in the first place. There was no way I was going to commission art work for a fucking Dungeons and Dragons mini-campaign. I won't even pay AI to make these images for me. Since I only use the free account. ---> Free account: is AI actually profitable? Does co-pilot make money, does chatGPT make money, does AI in business, what is its purpose? - to usurp and replace employees. Especially white collar. Middle class employees. This would increase unemployment and lower consumer spending as people react accordingly. So who is going to buy the business's products? Especially which the majority of businesses offering luxury services and products that are not necessities. ==> bad for the economy. - calude code by Anthropic makes AI code --> who the fuck is going to read it all and deal with the technical debts involved with having an undocumented codebase. The more you use AI to do your job for you, the more dependent you will become on it. Profitability issues https://www.youtube.com/watch?v=OunJtLnyPT4 https://www.youtube.com/watch?v=WPqoCJK8vBo https://www.youtube.com/watch?v=B07mYv-IeU8 --> too many loans https://www.youtube.com/watch?v=zdsoe_OsnHw --> AI is in a bubble spending investor money for products that doent make any https://youtu.be/zdsoe_OsnHw?t=636 --> AI cant be trained on AI output --> info degeneration (information inbreeding and mutation) "recursive training" +> who knew that drinking your own piss will eventually poison you. No Bear Grills XD https://www.youtube.com/shorts/KzjHNSRgHJ8 --> AI in propaganda "You are not immune to propaganda" Is video and picture evidence still considered as iron clad as it was 10 years ago in 2016? Negative sentiment w/ AI advertising https://www.youtube.com/watch?v=Yy6fByUmPuE https://www.youtube.com/watch?v=eWnP9f-ceqE privacy concerns https://www.youtube.com/watch?v=dB6pOolO7io cutting the User out of the workflow --> also means that you don't know what you are signing your name to https://www.youtube.com/watch?v=-vvGAKtV_Ek AI adverts https://www.youtube.com/watch?v=2l_2k5KrXmo https://www.youtube.com/watch?v=Xh2nsY9tQzs https://www.youtube.com/watch?v=qL7kr1ckgDE --> lazy idiots -> Asking the tool to read for them +> offloading cognitive labour, keeps your brain dumb https://www.youtube.com/watch?v=5cGAGr73Fkg - meaningless guff ad