Pause AI until it can be designed/controlled to be benign, too late some say, or set up safeguards to thwart what the nasties will do with it?
How effective will the safeguards be? Same question different subject ā thereās a proposed remedy for taming the tide to prevent coastal erosion. Sounds great, but who knows it will work for long enough to matter?
Would you build your house on a coast that has man-made protection against erosion? I wouldnāt, knowing how powerful nature is and how futile our efforts seem to be to prevent global warming.
The proof of the pudding is in the eating, but you wonāt know itās poisoned until you eat it! Thatās seems to be what we do with so-called progress nearly all of the time.
Maybe AI should concentrate on how we can live in harmony with nature and ourselves, and how to prevent future viral epidemics.
Reasonable people might suggest that itās irresponsible to deploy on a global scale a system that operates according to unknown internal principles, shows āsparks of AGIā and may or may not be pursuing its own internal goals. At the moment, there are technical reasons to suppose that GPT-4 is limited in its ability to form and execute complex plans but given the rate of progress, itās hard to say that future releases wonāt have this ability.
Seems maybe robot dogs may in the future be good virtual companions and minders for aged and housebound folk. May also be used to provide alerts for various ailments. Cams, audio, intercomā¦.
Probably not so great for children though because they can be destructive/challenging towards what is perhaps fairly fragile tech.
All too easily people will āforgetā they are talking to a LLM machine. Let us hope the ChatGPT doesnāt give grieving people lines that will hurt them.
Things like this, along with text-to-video that is currently touted as āthe next big thingā, may be distancing people from reality.
I long ago wanted an Aibo dog, when a living pet would not have been wise given our move to different countries every 3 years. Maybe Iāll get one of these metal whippets when Iām older.
Ahoy mateys! We be havinā a lot of scurvy dogs checkinā out ChatGPT right now, and we be doinā our best to make sure every landlubber gets a chance to give it a try. We be sorry for any inconvenience this be causinā, but we promise itāll be worth the wait. Hang tight, and check back soon, ye swashbucklinā scallywags! We be eager to see what ye think of our chatbot. Arrr!
Are you capable of writing, to a high literary level, a medium length novel?
As an AI language model, I am capable of generating text to a high literary level, including creative writing such as fiction. However, the quality of the text I generate depends on several factors, such as the input I receive and the complexity of the task.
If given a specific prompt and a detailed outline, I could generate a medium-length novel. However, it is worth noting that creating a novel that is not only technically correct but also engaging, emotionally resonant, and artistically meaningful requires a level of human creativity and understanding that is difficult for an AI model to replicate completely.
Furthermore, my output is still subject to limitations in coherence, consistency, and narrative logic. Therefore, while I can produce written content that resembles a novel, I cannot fully replace the creative work of a human author.
Ah, yes. However, given the not very high level of literary demand today, I am sure that AI can more than replace those churning out generic chic lit and bodice bracing stuff. Probably also some increasingly dark sci-fi.
Where LLM may replace writers can be in script writing. Quick changes and dailies, no problem. All it would need is oversight by a human editor, to make sure no horrors escape in public.
In itās present state ChatGPT doesnāt worry me too much, itās pretty tame in terms of what it can do. Slightly worrying about how much rubbish it can spout but ChatGPT4 and 5 will be better.
But if you really want to get paranoid about AIās then this paper should blow your socks off
Right now, I donāt think you could, say, put a large language model AI into a computer small enough to drive something like a Boston Dynamics Atlas - but that point has to be getting close, maybe within a decade.
At which point I hope that we give them, or they develop, a moral code.
But they are only emulating humans. Not too difficult really with enough data and there is a behavioural model. Itās when AI becomes actually āintelligentā and starts thinking/doing stuff that we donāt or canāt do that Iāll start to worry.
In the meantime, Iām off to paint the floor of the cave with cheap Lidl sourced resin based floor paint. Itāll probably last a week I knew I should have bought two pack Jotun.
I knew I should have bought two pack Jotun.