Thursday, October 22, 2015

Working towards a thinking machine




Artificial intelligence won't come easy.

Maybe I got overconfident with Watson. Maybe I bought into Kurzweil a bit too much, even though I still find him the most compelling writer and speaker there is on the subject. Or maybe I'm subconsciously being influenced by the Pandora-like criers, such as Stephen Hawking and Elon Musk, both warning us against what an age of intelligent machines might mean. I began to find doubt creeping into my assessment of artificial intelligence and if we would ever hit the 2030 or even 2050 mark.

We are, however, making great progress with developing computers that can learn. These machines assimilate information from raw data and sometimes can relate it back to us through voice programs. Think Siri only it actually works. The BBC recently reported on the work of Fei-Fei Li. Her idea was to develop computers that can see. If shown an image, the artificial intelligence program is then able to accurately label them. One example at the link shows that the algorithm has assigned the text "black and white dog jumps over bar" to a photograph that depicts just that.

It isn't exactly 100% but then what is? One photograph of a small baby holding a toothbrush was labeled "a young boy holds a baseball bat." The AI also doesn't do so well with abstract art.

The BBC link also has a really cool timeline of AI development up until this point. You can see the initial accomplishments of Alan Turing and Christopher Strachey all the way through to Google Now and Cortana. In light of that, I suppose my recent grumbling doubts are misplaced. It's a marathon, not a sprint. We're going to have fully realized AI. The exact date of it is what's up for debate.

Are you scared? No reason to be. They're not going to take over. I think that the first AIs will see that the real profit is in mass media. Seeing just how zombified we become with our smartphones, AIs will compete to produce the ultimate, immersive virtual reality experience. You say you're a TV junkie? Well you ain't seen nothing yet. America's about to become "AI and its 300 million bitches." You won't want to leave the show.

I also like my friend David's idea. After absorbing the sum total of human knowledge, a self-aware AI might simply utter "There's no point in anything" and then shut down.



Like ESE on Facebook




Follow me on Twitter: @Jntweets  

No comments:

Post a Comment