Being AI Pacino – AI standing for artificially intelligent – I can get away by making ‘interesting commentary’ on matches in the ongoing World Cup; make people turn towards me just by asking, ‘But won’t the latest non-manufacturing PMI affect the next Fed rate hike?’; throw in some unconfirmable Patidar poll behaviour to explain Thursday’s Gujarat poll results; and recommend, without giving too much away (since I have little to give away in the first place), Kotaro Isaka’s novel, Bullet Train.
After all, what is intelligence but artificially imbibing, cherrypicking, collating and then laying things generally accepted as ‘intelligent’ liver pate thick for folks to wah, wah? To be deft at pretending intelligence is to come across successfully as being intelligent.
Mathematician and logician Alan Turing would have recognised this as mimicking intelligence. Or would he have? Turing devised an experiment – the Turing Test, originally called ‘the imitation game’ — which set a standard for machines to be called ‘intelligent’. By asking a set of questions like a visa officer questioning a Rohingya refugee pretending to be an investment banker, the Turing Test hopes to catch an artificially intelligent machine trying to slip through as human, no matter how low the latter’s IQ may be. Or, at least, that is its plan.
Yes, some might call this racist against robots. But advancements in pretending have grown leaps and bounds. Everyone’s agog with news about the new Al in town – OpenAI’s ChatGPT (generative pre-trained transformer). When given (almost) any cognitive task, it can do it in seconds with aplomb. From solving your JEE maths problems on command, to writing an epic poem in the style of Chetan Bhagat on the subject of a Hindu invasion of Kentucky, to drawing up export projections of Jharkhand coal to Newcastle, ChatGPT can wow people who stopped getting wowed by GPS locating a while ago.
But before you can heehaw, ‘C’mon AI Pacino, you’re just jealous of this new highly capable linguistic brain,’ let me just say why ChatGPT will be a novelty item for 99.9% to generate jokes, memes, wedding invitations, (inadequate) course work, poetry (for people who can’t tell the difference between a Rimbaud and a cousin’s poem on a rainbow)…. In other words, a performing monkey for spectating primates.
After the initial wowzee dies down, (intelligent) people will find out that instead of replacing employees with ChatGPT, there’ll be five boffins for every terminal with ChatGPT, which may be cut down to two. (‘Sharmaji, how shall I frame the command to ChatGPT about coming up with a super-engaging ad copy for this mutual fund again?’) Unlike ATMs with their basic command choices, or sensor-activated taps with no command choices needed, the real value in ChatGPT operations is provided by the command-er. It’s how one frames one’s ChatGPT ‘ask’ that gives this AI its je ne sais quoi (an intelligent way of saying ‘USP’).
Till then, it’ll be like having all those airport personnel next to those boarding pass printing machines helping passengers to get their boarding passes printed. And by that time, ChatGPT or its offspring will be as dated and twee as shorthand and typing learning centres.
ChatGPT, at least in its current beta stage, already has a far superior alternative. By the very fact that no one has been able to spot that it is artificially intelligent makes it the brainpower to reckon with. That is until now, as I share the fact that this column, since its inception, has been written by AI Pacino, an AI program that is certainly more intelligent than the person whose name the column bears.