Robots Weekly ?: What Will AI Do Next?
There’s a lot of talk about what Artificial Intelligence will do in the Future.
Take all our jobs. Turn us all into paperclips. Or turn Earth into “strawberry fields forever”. To name a few.
AI is also the new buzzword that gets added to everything. So, in honor of all of this, let’s take a look at what AI will do next (or has recently done).
Break CAPTCHAs ?
Good news: you probably won’t have to deal with those annoying distorted squiggly letter humanity verifiers called CAPTCHAs for too much longer
Bad news: it’s because they built an AI that can turn those suckers out. An AI that is simpler and smaller than state of the art ones and used a smaller training set. So maybe be careful on the internet, which, honestly, you should be doing anyway.
(I think CAPTCHAs and that Google “identify parts of this picture” security thing are used to label training sets for machine learning systems, so now our AI can be used to generate data to build our other AIs?)
Go To College ?
Some researchers in Japan wanted to see if an AI could get into the University of Tokyo. So they built a robot to take the very rigorous entrance exam. Spoiler alert: it crushed it, ranking in the top 20%.
I think Avinash summed it up pretty well so I’ll just quote him:
“Among many implications from this progress, there is one that stands above and beyond all else: If our education system is geared towards teaching rote memorization, to optimize recall rates, paraphrase long texts, and be good at lots of subjects being taught in school (but exceptional at none of them)… What are we really solving for?
An AI is already better at this than most of us. Soon, it will be better at this than the best of us. In the very near future, an AI will be better at this than all of us combined.”
The future of space is smart. Smart satellites that is. The University of Akron in Ohio received a grant from NASA to research smartening up satellites with AI to help them prevent themselves from getting hit by space rocks or to get some sweet spur-of-the-moment research when the opportunity presents itself. All to counteract the communication time lag as these hunks of metal get farther from Earth.
Oh yeah, they also plan on using the blockchain. (Warning: unverified opinion to follow) I’m not sure what this part adds the equation. I think the blockchain has some serious disruption potential in all kinds of arenas and could be “the future”, but my knee jerk reaction is to think this is jargon stuffing to appeal to and excite the people in charge of the grant. Like how it will use “Ethereum”, the Hansel of crypto (so hot right now).
Experience Self-Doubt ¯\_(ツ)_/¯
Not only will AI overthrow humanity, but it will steal our standing as the only species to experience crippling self-doubt?! Outrageous! Seriously though, the Googs and the Ubes and working on adding the ability to handle probability to “the two most popular deep-learning frameworks”. This will help machines (read: self-driving cars) make decisions by moving beyond a binary yes-no to a level of confidence. A key factor of human intelligence.
Enter all the things!!! ?
AI will likely start to be baked in to all consumer electronics products. Phones, headphones, IoT devices, laptops, watches, glasses, stuff we don’t know about yet, and on and on. Google basically said as much last year.