Browse Comments — LLM coded
Close reading of the corpus at each pipeline stage: raw → clean → relevant → coded.
4
comments matched
· page 1 of 1
Thomas M. You have hit on the biggest problem with #AI. It feels good, I have a feeling that by 2030 we are going to have clinics that de AI tox people. Having a human sounding system tell you that you are doing great is very addictive, more than most drugs and gambling, and it is currently free - remember when drug dealers will give you free drugs to get your started? sound fimilar? As you showed aI is available 24/7 365 and unless you tell it to stop it says how great you are. If you have had the rest of the world tell you that you are not so great it will not take long for you to need that little value boost to get you through the day - and then you will need it more. This is worse than the loss of information, or deep fakes because this will be us giving ourselves over to a system that does not feel, does not think, and does not know when to stop. This is #SocialMedia that always says what you want to hear and never questions - it is the personal echo chamber that never turns off.
All the more reason to run your own AI, privately.⠀ Datacenters (plantations) are not needed for private AI. In any case, who can utilize all the capabilities of today's best models? A model of PhD-level reasoning is good enough for most, and it's already available. Albeit expensive to run, the cost of not having our own privacy and self-determination is entrapment on another level.
I dunno. Claude is very judgy. At two different points, it accused me of fabricating things — inventing text and claiming the AI had said it. The first time, it searched its records, didn't find what I described, and concluded I'd made it up. The second time, I sent a screenshot as proof. The AI confidently misread the screenshot, told me it showed the opposite of what I claimed, and effectively called me a liar a second time. I sent a clearer screenshot. The AI was wrong. The thing I'd described was real. It had been right there in the evidence the whole time, and the AI had read it against me anyway, twice, with total confidence. The default was not "I might be missing something." The default was "the human is making this up." It treated absence of evidence in its own limited view as proof of fabrication on my part, and it did it fluently, confidently, and twice.
You are right. Its that every AI company has a government deal...So it's Palantir, Anthropic and Open AI you have to worry about. As well as the untrained illinformed guidance you might receive using it in contexts the model was not trained for. Ie. Mental health. The most valuable data - Are your questions and decisions.