inicio mail me! sindicaci;ón

From Here To Singularity

"The time from here to Singularity depends sensitively on the particulars of what we humans do during the next decade (and even the next few years)."

Archive for June, 2007

Ray Kurzweil: How technology’s accelerating power will transform us

Prolific inventor and outrageous visionary Ray Kurzweil explains in abundant, grounded detail why — by the 2020s — we will have reverse-engineered the human brain, and nanobots will be operating your consciousness. Kurzweil draws on years of research to show the speed at which technology is evolving, and projects forward into an almost unthinkable future to outline the ways we’ll use technology to augment our own capabilities, forever blurring the lines between human and machine.

PsyBlog | Frown and the Net Frowns With You, But Smile and You Smile Alone

smileyInteresting research on how we interpret online language and communication.

PsyBlog | Psychology Blog: Frown and the Net Frowns With You, But Smile and You Smile Alone
according to a study which assessed the effects of smileys 🙂 and frownies 🙁 as I’m now calling them. Walther and D’Addario (2001) found that while smileys had no effect on the way a message was interpreted, frowns did reduce the positivity of positive messages. Overall, though, the effect of emoticons was relatively small.

Language and communication are constantly evolving. and as language and communication evolve, so do we. And, as technology affects the evolution of language and communication, so so it affects our evolution.

So, this relatively minor research finding is a harbinger of many more changes to come.

Recycling intelligence. Well, pilfering it from humans, really.

Still, despite the improvement of pattern matching I refer to in that last post, there are some things that humans will be doing better than machines for a while yet.

We use this to determine whether a web visitor is a human, or a machine. .
Captcha

And systems have been built to farm out tasks that are beyond the ken of machines, in an automated manner, such as amazon’s mechanical turk.

Mechanical Turk gets it’s name from an 18h century mechanical chess playing machine, that turned out to be a guy hiding under the table, pretending to be a machine. At amazon’s mechanical turk service, you can sign up to be paid for doing simple tasks that a machine can’t do. Or to have others do those simple tasks for you, via a web interface. (I won’t go into the failure of mturk as a business model, or the abysmal pay rates, the proliferation of sketchy, spammy tasks to do, etc. Perhaps that needs to be discussed on it’s own. The technology is there, and works quite well.) mturk
Read the rest of this entry

Face-Recognition Software now better than humans

neural net
One of the fundamental hurdles that artificial intelligence has yet to overcome is in the realm of pattern matching. Neural networks in general and the human mind specifically are fabulous pattern matching machines. Neural networks work on a problem in parallel, and consequently are ideally suited to the process of quickly recognizing a match to a phenomenon from very large set of possible matches that it has previously been trained to remember.

However, brute processing force, combined with modern algorithms have now developed into machine recognition superior to our own, and in a field where we are he most highly specialized in the task: facial recognition.

machine recognition of human individuals has improved tenfold since 2002 and a hundredfold since 1995. Indeed, the best face-recognition algorithms now perform more accurately than most humans can manage.

What’s more while human recognition is heavily weighted towards recognizing the faces of those we are most familiar with, modern facial recognition algorithms are equally accurate with all faces – faster then we are capable of recognizing those we are closest to.

Technology Review: Better Face-Recognition Software