NAB
“If we can’t understand how a
technology works, we risk succumbing to magical thinking,” says Jaron Lanier in
the tech guru’s latest contribution to the debate on AI.
article here
“Is there a way to explain AI that
isn’t in terms suggesting human obsolescence or replacement? If we can talk
about our technology in a different way, maybe a better path to bringing it
into society will appear.”
This is the week in which Elon Musk filed a lawsuit against OpenAI for putting profit before the good of humanity in proceeding full steam to develop next-level Artificial General Intelligence.
Since Microsoft has invested heavily
in OpenAI to do that, Lanier’s intervention could be seen as an attack on Musk
for raising the ‘threat’ levels of AI but Lanier is too smart and liberal an
operator for that.
He also wants all AI leaders to open
the black box and show us what is inside.
Lanier attempts to this in his essay published
in The New Yorker. He is alarmed by the fever pitched discussion of
AI where the loudest voices appear to be at its extremes. Those doomsayers
fearful of the technology’s inevitable human apocalypse and those who think
that humans will always be masters of their own destiny evolving with AI as a
force for overall good.
Some hold both positions.
“I have trouble understanding why
some of my colleagues say that what they are doing might lead to human
extinction, and yet argue that it is still worth doing,” Lanier writes. “It is
hard to comprehend this way of talking without wondering whether AI is becoming
a new kind of religion.”
Lanier advocates a third way, a
middle way, which he hopes offers an alternative to the view that AI does
nothing but regurgitate — while also communicating skepticism about whether AI
will become a transcendent, unlimited form of intelligence.
He thinks that we should start by
demystifying what AI is.
“We usually prefer to treat AI
systems as giant impenetrable continuities. Perhaps, to some degree, there’s a
resistance to demystifying what we do because we want to approach it
mystically,” he argues.
He continues, “One problem with the
usual anthropomorphic narratives about AI is that they don’t nurture our
intuitions about its weaknesses. As a result, our discussions about the
technology tend to involve confrontations between extremes: there are enthusiasts
who think that we’re building a cosmically big brain that will solve all our
problems or wipe us out, and skeptics who don’t see much value in AI.”
He takes issue with the term
“artificial intelligence,” suggesting it permeates the idea that we are making
new creatures instead of new tools. “This notion is furthered by biological
terms like ‘neurons’ and ‘neural networks,’ and by anthropomorphizing ones like
‘learning’ or ‘training,’ which computer scientists use all the time.”
It’s also a problem that “AI” has no
fixed definition.
“It’s always possible to dismiss any
specific commentary about AI for not addressing some other potential definition
of it,” he says.
The lack of mooring for the term
coincides with a “metaphysical sensibility” according to which the human
framework will soon be transcended.
In an
earlier essay he discussed reconsidering AI
as a form of human collaboration. Here he deconstructs how AI works for the
layman.
“Most non-technical people can
comprehend a thorny abstraction better once it’s been broken into concrete
pieces you can tell stories about, but that can be a hard sell in the
computer-science world,” he says.
The science-fiction writer Arthur C.
Clarke famously stated that a sufficiently advanced technology is
indistinguishable from magic. Lanier says that is only true if that technology
is not explained well enough.
He adds, “It is the responsibility of
technologists to make sure their offerings are not taken as magic.”
No comments:
Post a Comment