Quote:
Originally Posted by Irace86.2.0
Watson was able to beat Ken Jennings at Jeopardy because it was able to interpret play on words and the nuance in human expression from sayings to innuendos. Your tranny definition would be easy because of word association, not different than what you have to do. The AI could error on deleting posts or on leaving posts until things were obvious or until people reported the posts. ChatGP wouldn't be able to write a paper if it couldn't discern the difference between common and slang definitions. AI has bested humans in chess, Go, Jeopardy, has bested doctors in diagnosing diseases and so on. I don't expect to see a product for the forum anytime super soon, but it isn't far away either.
I think your bigger issue isn't that you will be replaced, but before that, it will become increasingly difficult discerning bots from real people. They have gotten a lot better over the years. Nefarious links are easy for an AI to test and discard, but something like a scammer, that would be harder to identify and remove. Imagine an AI selling a product with a perfect CGI picture of something fake, and then getting multiple people to buy their fake product. Or something else.
|
I get your pointing of Watson winning the Jeopardy… even though failed at the trivial questions, especially final question was an interesting. I would have expected trivial should be an AI’s expert topic.
AI can analyze data & may understand to certain degree (including the sentence), but they cannot critical thinking or analyze data/sentences as human can (atm). Just like AI cannot define “beauty”. AI can objectively look or read, but cannot subjectively look or read. I guess this can be said the same for human, especially in the forum or text form. Discussion is made 93% non-verbal 7% verbal. Within 93% intonation & body language are the majority of the factor. This is why miscommunication between human happen online than face-to-face. Cuz they misread or misinterpreted the sentence (depends on person’s mood).
I’m not really worry about AI selling product with perfect CGI pics of something fake. Unless it’s simple shift knob or other small parts, majority of buyer PM the seller w complicated, random question even human have hard time answering or some are pure WTF questions. If AI become that quick-wits & unique enough to answer convo like that, then prob be used for seniors &/or other solo living ppl first… but of course, spammer/hacker will use for their greed before that.