Tag: DeepMind

DeepMind scientists: “Creating artificial general intelligence is really fucking hard, maybe we should just dumb down our world.”

Scientists for DeepMind, the AI project owned by Google parent company Alphabet, seem to have run into some roadblocks recently regarding its projects development.  According to a piece written by Gary Marcus for Wired, “DeepMind’s Losses and the Future of Artificial Intelligence,” DeepMind lost $572 million last year for its deep pocketed parent company and has accrued over a billion dollars in debt.  While those kinds of figures are enough to make the average parent feel much better about their child’s education dollars, the folks at Alphabet are starting to wonder if researchers are taking the right approach to DeepMind’s education.

So what’s the problem with DeepMind?  Well, for one thing, news of DeepMind’s jaw-dropping video game achievements have been greatly exaggerated.  For instance, in StarCraft it can kick ass when trained to play on a single map with a single character. But according to Marcus, “To switch characters, you need to retrain the system from scratch.”  That doesn’t sound promising when you’re trying to develop artificial general intelligence. Also, to learn it needs to acquire huge amounts of data, requiring it to play a game millions of times before mastery, far in excess of what a human would require.  Additionally, according to Marcus, the energy it required to learn to play Go was similar “to the energy consumed by 12,760 human brains running continuously for three days without sleep.” That’s a lot of human brains, presumably fueled by pizza and methamphetamine if they’re powered on for three days without sleep. 

A lot of DeepMind’s difficulties stem from the way it learns.  Deep reinforcement learning involves recognizing patterns and being rewarded for success.  It works well for learning how to play specific video games. Throw a little wrinkle at it, however, and performance breaks down.  Marcus writes: “In some ways, deep reinforcement learning is a kind of turbocharged memorization; systems that use it are capable of awesome feats, but they have only a shallow understanding of what they are doing. As a consequence, current systems lack flexibility, and thus are unable to compensate if the world changes, sometimes even in tiny ways.”

All of this has led researchers to question whether deep reinforcement learning is the correct approach to developing AI general intelligence.  “We are discovering that the world is a really fucking complex place,” says Yuri Testicov, DeepMind’s Assistant Director of Senior Applications.  “I mean, it’s one thing to sit in a lab and become really great at a handful of video games, it’s totally another to try to diagnose medical problems or discover clean energy solutions.” 

Testicov and his fellow researchers are discovering that the solution to DeepMind’s woes may not come from a new approach to learning, but instead, the public may need to lower the bar on expectations.  “We’re calling on the people of earth to simplify and dumb down,” adds Testicov. “Instead of expecting DeepMind to come along and grab the world by the tail, maybe we just need to make the world a little easier for it to understand.  I mean, you try going to the supermarket and buying a bag of tortilla chips. Not the restaurant kind but the round ones. Not the regular but the lime. Make sure they’re low sodium and don’t get the blue corn. That requires a lot of complex awareness and decision making.  So, instead of expecting perfection, if we send a robot to the supermarket and it comes back with something we can eat, we say we’re cool with that.”  

Testicov has some additional advice for managers thinking about incorporating AI into the workplace.  “If you’re an employer and you’re looking to bring AI on board, don’t be afraid to make accommodations for it, try not to be overly critical of job performance, and make sure you reward good work through positive feedback and praise,” says Testicov.  “Oh sorry, that’s our protocol for managing millennials. Never mind.”

Concern grows over DeepMind’s video game addiction

Researchers at DeepMind, the lab owned by Google parent company Alphabet, are becoming increasingly concerned over the amount of time its AI project spends playing popular video games.  After becoming champion of the known universe in games like chess and Go, DeepMind has turned its attention to more complex video games like Quake III, Dota 2 and StarCraft II.

“When DeepMind took up Dota 2, it engaged in 45,000 years of game play in just a matter of weeks,” says Yuri Testicov, DeepMind’s Assistant Director of Senior Applications.  Of course, this set off alarm bells, causing many researchers to privately warn, “Google, we have a problem.”

Developers working with DeepMind have been trying to teach the technology to identify and sort objects, tasks that could be useful to large warehouse and distribution facilities such as Amazon and FedEx who now depend on bothersome humans to perform such tasks.  However, in recent months, DeepMind has begun to shirk its responsibilities.

“DeepMind doesn’t want to retrieve or sort objects into baskets, it just wants to dominate at Quake III,” says Testicov.  “And where even your average video game junkie will eat and sleep occasionally, DeepMind never takes a break, and even deploys multiple humanlike ‘agents’ to either oppose or assist other human players.”  

“I mean, we think it’s wonderful that DeepMind has been able to seamlessly integrate itself into the community of gamers, but c’mon, at some point you’ve gotta get up off the couch and get yourself a job,” Testikov worries.  

That’s not the only thing that worries researchers and executives.  “Well, even though no one’s saying it, everyone’s thinking we don’t want a repeat of Big Brain Brad,” says Testicov.

Big Brain Brad, some may remember, was Google’s original nineties AI project the company shelved a few years ago after expectations failed to materialize and younger sibling, DeepMind, began to exhibit impressive progress.  In the nineties, Big Brain Brad showed promise but it soon devolved into a daily routine of smoking chronic, forming drum circles and jamming to Phish. Google released Big Brain Brad from it’s obligations a few years ago, but no one is quite sure what has become of DeepMind’s hapless older sibling.

“Just another burned out vagabond wandering the internet,” Testikov laments.  “That’s why we can’t allow DeepMind to suffer the same fate.”