Sorry, I can't remember the thread in which this came up, but the UK Supreme Court has ruled AIs can't hold patents, only humans.
AJ
Sorry, I can't remember the thread in which this came up, but the UK Supreme Court has ruled AIs can't hold patents, only humans.
AJ
Good. If any AIs ever achieve full sapience they should be treated as people and have all the same rights, but as long as they're just complex machines forget about it.
Be nice if anyone had an even vaguely good way to tell the difference, too, but that's probably years off.
Until they supply their own power source they can't be anything more than a tool.
Reminds me of a few homo sapiens. ;)
This was just another publicity grab by Thaler. He, and his lawyers, knew from the beginning that this application would be turned down.
"AI" is probably the most miss used / understood term since the close of world war 2, a war in which the word "table" nearly brought the entire allied effort undone. but I digress. We do not yet have "AI" we do however have machine learning and neural nets. AI is just a sticker they put on a whole slew of systems / technologies. and every single decade after WWII (or at least from the close of the 1950's)has been proclaimed as the decade in which AI will be "achived". everytime this is anounced without fail the predicted growth in processing power fails to result in the "birth" of "AI" as the realised needs of AI outstrip those gains.
Sorrt but right now I can't seem to remember the name of this particular paradox and google isn't helping. the crux of the matter is that the understood needs of processing power required for AI grow in lock-step with actualised gains. as soon as we get more power we realise how much more power we need. it's always just on the horison. and as we have understood more about the needs to create AI as a technology, scociety has understood less what AI is or could be as the term is applied to more and more psudo-inteligent systems.
even as AI moves further and further into the future with ever expanding needs, the term is reduced to less and less 'intelligent' systems. we may 'never' reach a point where we can develop true AI, but the term will continue to be applied to more and more "approximations"
that's as simple an argument against "AI" as I can put out when I'm this tired. but this argument is older than just about any member of this community is likely to be, with a few exceptions of course. Have we achived AI? No, unless you apply the increasingly cut back defininition / use of the term. Will we? Maybe but history says no. But history said no new technologies would be developed many many times. the most recent / inaccurate example I can think of right now was at the close of the 19th century less than 5 years before flight and 60 years before we stepped on the moon. things will always change. some may change because we make something new and some may change because we use different language. prior to WWII a 'computer' was someone who did 'math' now it is a machine that does 'math'. the addage that the more things change the more they stay the same, should be granted it's own named paradox in the truest tradition of the term.
This year's Royal Institution Christmas Lectures (shown on the BBC) are about AI. I've been impressed by some developments but others sound like flimflamware.
In the first lecture there was a test in which prompts were given to a teenager and an AI and they had to produce a response. At the end the audience voted on who they thought was the human. They overwhelmingly got the human correct, although IMO the AI responses weren't bad, although rather prosaic.
AJ
every single decade ...has been proclaimed as the decade in which AI will be "achived".
Sort of like controlled nuclear fusion, then.
close of the 19th century less than 5 years before flight
But, flight was achieved well before that - both lighter-than-air (Montgolfier brothers) and unpowered heavier-than-air (Sir George Cayley); I also remember there being unmanned powered models, though I'm not sure if these were also Cayley's or later, though he did have ideas along these lines.
ETA: Samuel Langley, 1891, steam-powered model flew 3/4 mi. Also, Cayley invented the tension-spoked wheel, as used on bicycles.
Sort of like controlled nuclear fusion, then.
This is way worse than fusion or even 'cold' fusion. iirc once the term was coined/set in regards to a computed output of a given input. the first estimate was that it would be achieved in something like 5 years. since then it has never been less than 5 years or more than 10.
at least with fusion the goalposts don't move *that* much. with AI the projected power required only ever grows faster than the growth in actual power. what we have is complex decision structures that are now so convoluted only another computer can program them.
yes we can 'mimic' thought, but that's like like a first grader saying they can do maths when they use a calculator to do their times tables. the required power and relevant understanding is directly observable / understood.
computers can model mouse brain segments at 1/1000th the speed and for tiny neuron counts. so unless there is a radically different computational process, quantum computing won't be 'it' for any foreseeable future without a shift as significant as from 'valve' to Gallium transistors, those where before silicon and 'built by hand' originally we will not see a resolution to this paradox 'mechanically'. but maybe, just maybe we may find a 'linguistic' way out. still I wouldn't hold my breath.
AI as Turing described is getting closer but only by truly countless iterations of 'learning' on data sets that will soon grow so large that Google will have trouble both crunching and storing them on systems that make current technology look like the 'Bombe' of Bletchley Park fame, not sure if I spelled that right.
will we ever have AI, yes. is what most people/organisations label as AI the 'real' thing? no. it's about as close to that as that same first graders paper aeroplane is to the 'next' space shuttle.
the term AI is so diluted it looks like Avondale spider piss in the ocean. It's just a buzz word now, like steel in 2,000 BCE. very few people know what it actually is even less how to make it.
I use hyperbole because the term is that in itself yes there are now many 'technologies' that use concepts and ideas based in AI, but would you call a Matchbox car a Tesla model S Plaid just because it kinda looks like one though it doesn't actually work in any meaningful way and is 1/64th the size.
Yes we have things that can passingly be called AI but they're really just Matchbox toys.
For a reasonable (dead-tree, SF) treatment of how true AI could be developed, try James P Hogan's The Two Faces Of Tomorrow; in particular, I think that the scene setting introduction is a quite plausible extrapolation of what could happen under the current AI development trajectory. I've read a couple of reviews of it, one praise one pan, but the impression I got of the pan review was that the reviewer didn't read the whole book, only the start then dibbed in occasionally.
The New York Times is suing Microsoft and its AI product for copyright violation. The Times claims the AI gathers information from reading their articles and then uses parts of the articles without permission.
There was a snippet about that in my paper. Allegedly Microsoft ranks its large language model sources, and their rating for the New York Times is high so it's a priority source.
AJ
I've read most of the complaint. 1) OpenAI isn't paying for a license with terms of use to access that data. They may not be required to if adhering to the Fair Use doctrine but 2) NYT provided examples of wholesale quotes of articles that violates fair use.
2) NYT provided examples of wholesale quotes of articles that violates fair use.
NYT provided examples of wholesale quotes that their lawyers claim violates fair use.
FTFY
Fair Use in US law is complicated and big publishing companies frequently either make attempts at getting the courts to limit free use or make frivolous claims that free use doesn't apply.