4 ms·
I don’t understand why people find this hard: chargpt is not sentient. None of them are. They are a bunch of glorified statistical modes. You type something, t
by olliej 4y ago
I don’t understand why people find this hard: chargpt is not sentient. None of them are.
They are a bunch of glorified statistical modes. You type something, they use a statistical model to decide what to look up, then they use a statistical model to weigh the importance/relevance of each thing they find, then they use a statistical model to make the answer text look like a human answer according to a statistical mode they built.
That’s why the happily produce incorrect answers “made to look real”: it’s just Google with a prose engine in front of it, and for some reason people now think webmd etc are more accurate because the response has been laundered through that engine.
- ojosilva 4y agoIt's not that simple. ChatGPT has taken us into this uncanny valley of sentience, and even smart tech journalist (cough) activists (cough) twitterists can be sucked into the media circus of spotting the HALs and Roy Battys hidden behind our beloved chatbots. Duh. Now, OP's got a point: advanced LLM-based tools are moving extremely fast into our daily toolchain, in creative and unexpected ways, with minimal moderation and no supervision. It's not about its failures being a reliable knowledge base, but at how keen some of us are willing to hand the keys to their bash prompt to an statistical "perplexed" system with rand() bolted on in more than a couple places.
- andrewmcwatters 4y agoI don't think anyone is debating that. I think people are debating, "How much are we just biological statistical models?" and then working backwards to understand at what threshold do we consider intelligent-like behavior sentient. When are we just creating something intelligent and "sentient" but incoherent?
- ethanbond 4y agoPeople find it hard because no one knows what sentience is, how it emerges, and how to test whether a thing (including humans!) other than yourself has it.
- KyeRussell 4y agoI am sick to death of this frankly pseudo intellectual take. It completely ignores the fact that a) we have no idea how consciousness works, and b) there are plenty of humans that say inaccurate things. Explaining how ML works doesn’t change its behaviours or its effects.
- Kim_Bruning 4y agoPut that way, pretty much everything in the known universe is a glorified statistics model in some form or other.