VR Pioneer Jaron Lanier

October 3, 2000 11:20 PM

VR Pioneer Jaron Lanier trashes AI and futurists and makes a case that software designers that engage in this groupthink produce user-unfriendly software.
posted by skallas (4 comments total)

While I agree that there are very few signs, after 20 years of research, that computers are anywhere near approaching true AI, I don't believe we'll never be able to program a computer to behave in an intelligent (cognitive) manner, I just don't think it's likely to happen in my lifetime, we don't understand the 'mechanics' of our own thought process well enough.
He's certainly right about Word, each version seems to have more and more intrusive elements coded in and (more annoyingly) installed enabled, requiring you to spend an age disabling all the disk-hogging guff that comes with it.
Imagine all the productivity gains if business returned to buying the tool for the job instead of 'upgrades' which are actually just a few bug fixes and 5Meg of 'assistants' 'wizards' and other stuff that nearly everyone turns off straightaway
posted by Markb at 2:45 AM on October 4, 2000

Who actually wants an AI interface? I hear Mr Jakob Nielsen talking about this all the time. When you've only given an ambiguous instruction AI usually means it will try and define what you want and do what it thinks is best from within the context. Better interface, yessum, AI is bad for anything but Aibo.
posted by holloway at 4:15 AM on October 4, 2000

Why AI? I'll tell you -- because you're a freakin' moron who can't be trusted with computers, so they'll have to make the computer think for you. The computer is perfect, and YOU are so severely flawed you're lucky we let you even LOOK at a computer.

Well, OK, so nobody says it that way. But that's what's behind most (not all) of this AI-in-the-home thinking, IMHO. The basic logic is "users are stupid because they can't figure out my fabulous software, therefore I must make the software even MORE fabulous so that it can read their minds, and then I won't have to listen to them at all because my program will do it for me."

Or maybe I'm just cynical about commercial AI & NLP efforts. They all just seem so condescending....

In the New Regime, all consumer-market AI efforts will be preceded by at least one complete interface overhaul. If that fails, the UI designers will be executed, and only then will programmers be permitted to attempt AI and/or NLP.
posted by aramaic at 6:07 AM on October 4, 2000

I'm hoping this helps seal the coffin on the consant stream of futurist predictions, which are always wrong, and get back to reality. Way too many CS types have fallen into the Lanier trap of "the computer above everything else" and have shown us the shitty outcome of their labor. I'm all for "self-learning" algorithms and AI research (most of which is 'intelligent' code fulfilling a predefined task in a suggested but not forced way), but this stuff doesn't belong anywhere near a desktop PC.

Go read some sci-fi if you're in love with AI, but my computers are tools and the less flakery thats inside of them the better it is for me.
posted by skallas at 10:32 AM on October 4, 2000