Do you follow Brian Roemmele, Hursty? If not give him a look, what he posts daily is very interesting in terms of the development. Its equal parts amazing and terrifying.

    LT42 I’ll check him out, cheers for the tip

      It this ever gets to the point where it can realistically match my level of message board BANTS then I will be astonished.

        mono-stereo
        Now that would be dangerous.

        Would probably have to regress it back to Smallman1.0, essentially the same bantzzz as the robot in Short Circuit

            Hursty What a time to be alive, who do you believe, the human or AI?

            If Mr Hood was spouting nonsense then it wouldn’t take long for his claim that he wasn’t actually jailed to be rubbished and he would have zero grounds to sue.

            Given what ChatGPT said seems to indeed to be slanderous, the question should rather be whether AI can be sued or not?

            Its fairly random and not uncommon propensity to lack accuracy is certainly of concern. No doubt there are already plenty of thickets out there accepting its answers uncritically in their entirety.

            A newspaper publishing the same kind of thing would certainly be liable for being sued.

            If he’s successful it could really open the floodgates and potentially suddenly turn the whole thing from a potential asset to a massive liability for its owners.

                hugopal I’m fairly certain the current laws don’t account for the ability to sue AI itself?

                It would be funny to see how well it defended itself given the chance, regardless of its inaccuracies it is supremely powerful in many respects

                    Mad_Cyril Would probably have to regress it back to Smallman1.0, essentially the same bantzzz as the robot in Short Circuit

                    I asked ChatGPT if Ed is at the bants top-table and it replied “Yes”!?!

                    Clearly the whole thing is deeply flawed and has a fondness for a small pool of repetitive, pre-programmed answers.

                    • LT42 replied to this.

                        Hursty hugopal I’m fairly certain the current laws don’t account for the ability to sue AI itself?

                        I didn’t mean it in the sense of the literal AI, but as in whether the creator, OpenAI, can be sued on the basis of being a publisher.

                            hugopal
                            Clues in the name I reckon, artificial Intelligence is only gonna be as accurate as the algorithm it’s based on.

                            Anyone who parrots it’s output as their own work should be made to reveal where their data is from, is that not the point of a citation?

                                hugopal yeah I assumed you mean’t that but again I’m not sure what the laws are surrounding it as essentially they probably only set the parameters for the algorithm to work within. The content and delivery is then learned by AI itself so the creator has no real way of knowing what it may be

                                    mono-stereo Anyone who parrots it’s output as their own work should be made to reveal where their data is from, is that not the point of a citation?

                                    Indeed - but then you’ve got the major issue that ChatGPT doesn’t publish its own citations. At least with wikipedia it clearly cites its original source for things.

                                    And where the heck did it get its false info for that Australian banker story from, for instance?!

                                        Hursty The content and delivery is then learned by AI itself so the creator has no real way of knowing what it may be

                                        At the very least they’re surely going to have to spend a lot of money on moderating its output - as the likes of twitter and facebook have to do with the content that users write on its platform.

                                            hugopal good point. It’s clearly not infallible, and I suspect it’s main use case will be bland administrative written communication, rather than changing the world.

                                              hugopal Content moderation from AI is going to be a very interesting field especially when it incredibly advanced, what fail safes are implemented if any?

                                              Its all too little too late, look at the crypto regulation, technology advances way too fast for governments to have any tangible control

                                                hugopal Clearly the whole thing is deeply flawed and has a fondness for a small pool of repetitive, pre-programmed answers.

                                                It’s now miles passed that stage, Hugo. Welcome to the party.