Perhaps you’ve seen news stories that come out every so often about how a computer AI has mastered some new game, proving it’s prowess by besting the best human champions our world has to offer. If you haven’t, hear’s a quick refresher.
What’s fascinating about this latest game falling to the processing power of machine’s neural network is the open-ended nature of the game. While games like Chess and Go feature a mindboggingly high number of move combinations and possibilities, all of those possibilities are known to the computer because all of the pieces dictating those possibilities are visible.
Not so for real-time strategy (RTS) games like StarCraft II. When the game’s “fog of war” keeps the opponent’s moves and tactics hidden, the computer must make strategic choices based on imperfect and incomplete information. When a computer-agent operating in an environment like that whips the best in the world 5 games to 0 … well then. Just listen to the defeated champion’s take:
“AlphaStar [the computer player/agent] takes well-known strategies and turns them on their head. The agent demonstrated strategies I hadn’t thought of before, which means there may still be new ways of playing the game that we haven’t fully explored yet.”
This kickoff article is the doorway to an entire issue of articles all exploring the world of artificial intelligence here at the beginning of 2019.
“we’re now at the point where AI is going to get normal fast. … so normal you won’t even notice.”
A team of researchers at Columbia University are working on a system that combines electrodes implanted in the brain to detect brain waves, a voice synthesizer, and a deep learning AI to make it all go. The goal is to be able to produce in computer generated speech the words that the test subject had heard by measuring the brain waves generated by the hearing, processing, and decoding of speech.
You can listen to the audio output of the effort yourself.
Yes, it’s been revealed in yet another way just how hungry Facebook is for the one asset they worship and crave above everything else: data. This time, it involved paying teens $20 per month for them to use a VPN app (with its connection to Facebook not at all clear) that basically served as an open conduit for Facebook to monitor nearly everything about the participants’ lives visible through their phone, texting, and web usage. But, it’s ok, says COO Sheyrl Sandberg, because they consented and were compensated. (Yes. She actually said that.)
Seeing these behaviors only through the lens of mobile phone apps distorts the picture so much we forget what it is we’re actually talking about. Imagine a flesh and blood stranger paying teens $20 to let the stranger follow them around wherever they went, and record everything they did … with $20 more for every friend the teen brought to the stranger.
Meanwhile, Facebook has written code into their product neutralizing the efforts of journalism outfit ProPublica to compile a database of political ads being run on Facebook. The whole point of ProPublica’s work was to provide transparency on the ads being run and how they are being targeted to Facebook users.
Facebook has put the kibosh on that work. They’re building their own tool for folks to use, they say, that will be as comprehensive and transparent for users as was ProPublica’s. Facebook wouldn’t lie about that to shield the workings of their money making machine, would they?