Hey @BBCNews, when you report that @LeicesterTigers defeated a “young” @ExeterChiefs team to win the @premrugby cup, I make the average ages of the matchday squads to be 25 and 24, respectively. ∴ both teams were “young”.
I played in & witnessed some incredible games of rugby but how good did @Scotlandteam play v @FranceRugby today? #6Ns
50 points for Scotland - France 40. A remarkable game of rugby. This 6 Nations just keeps giving. One of the all time great games in the history of the game!
Apple understands that users miss the experience of always inserting USB cables wrong the first time, and wondered how they could recreate this experience with USB-C…
What is becoming dangerous, however, is that the plausibility of AI’s excuses are many times better than Eliza’s. Give Eliza a whirl at https://t.co/7VizFyfJwi and see what you think. (7/7)
AI - From ELIZA to ChatGPT
In the early 1980s, I worked at BP, in a department called Technico-Economic Appraisal (TEA). I wrote a paper on AI, called something like “Artificial Intelligence - Artificial Certainly, but not Intelligent”. I reviewed its prospects (1/7)
Although I have stopped listening, I think Musk is still promising it for next year. There was a writer for Byte magazine in the 1980s who used the term “real soon now” about software promises which still seems to apply to many aspects of AI today (6/7)
So much so that its answers are, frankly, laughable. I find my experience with many AI supported chat support bots to be similar. Indeed, when Elon Musk first tried to sell me “Full Self Driving” capability in 2016, I declined for much the same reason. (5/7)
how AI could ever work well beyond its training sets.
Recently, I have been putting ChatGPT through its paces and it reminds me very much of Weizenbaum’s Eliza. In its (much larger) domain, ChatGPT works well but outside that it is hopeless. (4/7)
As far as I recall, my conclusion was that AI was worth further research to establish which areas might be yield early dividends. One of those were Expert Systems and that proved to be the area I moved into next. The problem then was that it was extremely difficult to see…(3/7)
including a (fairly old) natural language processing (NLP) computer program called Eliza, which had been developed by Joseph Weizenbaum at MIT. I remember running it on the DEC VAX or PDP-11 we used at the time.
( 2/7)
May I suggest, @OfficialFPL, that feigning injury should rank somewhere between raking an opponent’s achilles (yellow card) and pulling someone’s long hair (red card)? Personally, I think it worse than hair pulling, but that already has the greatest possible sanction.
Hey, @tim_cook, why does my locked @apple Watch tell me the wrong date until unlocked? It tells the correct time. It can be bad for my health as I keep thinking I have missed important dates.