So, IBM Has an AI (Artificial General Intelligence)
IBM is pitting Watson, a de facto artificial intelligence the size of an RV, with a library of 200 million pages of assimilated information, against the top two Jeopardy champions of all time. Whether or not their supercomputer -- using 2,500 processing cores capable of handling 33 million operations per second each -- actually wins, the very fact that an AI now exists which is capable of answering Jeopardy questions is a turning point in the quest for advanced, human-equivalent and human-superior artificial intelligence.
The key is ambiguity and specificity. Jeopardy "questions" are known for throwing in odd and offbeat humor and metaphors while demanding that contestants sort through that in an average of 3 seconds and come back with the specific answer required (phrased as a question). Whether Watson succeeds or fails, IBM will have produced more than a search engine, but an AI capable of sifting through vast data archives to draw out not just a meaningful answer, but the answer, in context, that the user is looking for. A user (Alex Trebek) who will apparently be interfacing by way of verbalized questions.
If this is possible, and clearly it is, Watson will provide governments and research centers the ability to sift data automatically, with an AI not merely capable of "running a search," as Google's algorithms accomplish, but thinking about the question and weighing potential answers -- all in the space of a second or less, and all without requiring a live human to oversee the process, but only its end result.
Consider this fact in the larger context. We already have two computers doing scientific research -- one which sifts through medical journals looking for secondary uses of pharmaceuticals and their analogues. The other can take the genome of a simple animal such as a nematode and sort through its genes one-by-one, conceiving, designing and performing experiments to test the properties of each one of thousands of genes, handling the process effectively all by itself until the task is done.
Meanwhile, we have Google's mighty search engine, a Droid app which can make reservations at some restaurants at the imprecise verbal command of its owner, and a host of publicly developed apps for the iPhone. What we have, in effect, is a rapidly evolving version of what the science-fiction game Eclipse Phase would call a "Muse," a basic artificial intelligence that is with its owner from its earliest years, and which handles essentially all of its meaningless digital "paperwork" and other tedious jobs.
But between the emergence of that technology and Watson, we have all the tremendous strides forward in between. Given that we already have computers doing real scientific work and now capable of searching the world's collective memory, those steps are apt to be substantial indeed. Already, a bit of crowdsourcing apps and off-the-shelf technology might make a primitive Muse possible, and of course computers are already adept at operating within the artificially limited parameters of the Web.
AI advancing at this pace will have tremendous impacts on employment, shifting career and business opportunities, technological development and other issues I will go into more extensively on this blog. But I leave you with one thought...
Formidable as this technology is, how much more potent does it become in the hands of extraordinarily gifted human beings, who are working to tap their own fullest potential in terms of not only the machines they use, but the skills they practice, the personal enhancements they embrace, and the lives they live?
The key is ambiguity and specificity. Jeopardy "questions" are known for throwing in odd and offbeat humor and metaphors while demanding that contestants sort through that in an average of 3 seconds and come back with the specific answer required (phrased as a question). Whether Watson succeeds or fails, IBM will have produced more than a search engine, but an AI capable of sifting through vast data archives to draw out not just a meaningful answer, but the answer, in context, that the user is looking for. A user (Alex Trebek) who will apparently be interfacing by way of verbalized questions.
If this is possible, and clearly it is, Watson will provide governments and research centers the ability to sift data automatically, with an AI not merely capable of "running a search," as Google's algorithms accomplish, but thinking about the question and weighing potential answers -- all in the space of a second or less, and all without requiring a live human to oversee the process, but only its end result.
Consider this fact in the larger context. We already have two computers doing scientific research -- one which sifts through medical journals looking for secondary uses of pharmaceuticals and their analogues. The other can take the genome of a simple animal such as a nematode and sort through its genes one-by-one, conceiving, designing and performing experiments to test the properties of each one of thousands of genes, handling the process effectively all by itself until the task is done.
Meanwhile, we have Google's mighty search engine, a Droid app which can make reservations at some restaurants at the imprecise verbal command of its owner, and a host of publicly developed apps for the iPhone. What we have, in effect, is a rapidly evolving version of what the science-fiction game Eclipse Phase would call a "Muse," a basic artificial intelligence that is with its owner from its earliest years, and which handles essentially all of its meaningless digital "paperwork" and other tedious jobs.
But between the emergence of that technology and Watson, we have all the tremendous strides forward in between. Given that we already have computers doing real scientific work and now capable of searching the world's collective memory, those steps are apt to be substantial indeed. Already, a bit of crowdsourcing apps and off-the-shelf technology might make a primitive Muse possible, and of course computers are already adept at operating within the artificially limited parameters of the Web.
AI advancing at this pace will have tremendous impacts on employment, shifting career and business opportunities, technological development and other issues I will go into more extensively on this blog. But I leave you with one thought...
Formidable as this technology is, how much more potent does it become in the hands of extraordinarily gifted human beings, who are working to tap their own fullest potential in terms of not only the machines they use, but the skills they practice, the personal enhancements they embrace, and the lives they live?
posted by Ralph Cerchione | Thursday, February 10, 2011
0 Comments:
Post a Comment
<< Home