I recently watched a somewhat unsettling but thought-provoking film called Ex Machina that explores some of the possibilities (great and terrifying) of Artificial Intelligence (AI). In reading up on the film, I came across a quote from the director emphasizing that, while this film takes place in the future, it’s really only “ten minutes from now.” In other words, as far fetched as a nearly-sentient robot may seem, we shouldn’t be too surprised if, say, tomorrow we find out that a tech company has successfully engineered a complex AI device.
AI is everywhere, not least culturally--just think of HAL from 2001: A Space Odyssey or WALL-E or BB-8 in the new Star Wars film--and we are compelled, it seems, to fashion our computers more and more after ourselves. That’s why we find R2D2 and C-3PO so charming--they share traits with us humans. But is there a limit? The Uncanny Valley hypothesis says that if the robot looks almost, but not exactly, like a human, most people will be repulsed, yet new research shows that giving a robot a face may still the best way to endear the machine to us.
However exciting and/or terrifying humanoid AI may be, this is not the predominant form AI is taking, at least not presently. In fact, the AI that have the largest share in the market are often working away without most of us even realizing it. Search engine algorithms, Amazon recommendations, photo recognition, real-time language translation: many of these technologies are the result not of purely human-written programs, but of machine learning. Machine learning--specifically the branch called deep learning--refers to computer programs that, through little to no human intervention, are able to employ observation, logic, analysis, and other skills to teach themselves and improve their performance, often at incredible rates. For more examples and a bit of background on deep learning, check out Jeremy Howard’s TED Talk called The wonderful and terrifying implications of computers that can learn.
This form of AI comes with it’s own set of issues, most notably “bot-sourcing.” With machines now being able to learn skills and tasks at a rate far faster than humans, many jobs in the future may be outsourced not to China or India, but to AI. As mentioned in the TED talk above--as well as in a number of otherplaces--AI has the potential to completely restructure our workforce. If machines eventually acquire enough skills to take on a large share of jobs, what does our labor force look like? It sure sounds nice to have a robot do all of our work, but transitioning to that world will likely be no easy task.
If nothing else, watching all of this unfold is going to be very interesting, and the more you know, the better. In addition to the films already mentioned, DPL has a lot of resources on AI, ranging from the dramatic and fictional to the instructional. We also have plenty of info about AI’s siblings: automation, electronics, and robotics. After all, why have a computer, if you aren’t going to put it to work? How much and what kind of work--well, that’s up to you.