FREEDOM AND SAFETY
Artificial intelligence is rapidly developing and is already starting to change the world, at a pace that is worrying to some experts. Huge personalities in the tech industry often lament the dangers of unfettered development of AI systems: people like Elon Musk and Stephen Hawking, who warn of a future where AI reigns supreme. Whether or not their concerns are unfounded, there certainly is a value on keeping tabs on the progress of AI innovation.
AI is getting good and in a lot of cases, way better than experts imagined. AlphaGo, Google’s game playing AI, has been beating the world’s best players for a while now, something that wasn’t thought to be possible for at least a decade. Elon Musk’s OpenAI is doing even better than that by beating the world’s greatest eSports players at DOTA 2, a game that is much more complex and involves tricking opponents. It is easy to see how an argument can be made saying that we need to more closely and accurately map the development of AI systems.
There are now multiple efforts to track AI and analyze its current and potential impact. One, called the AI Index, is being headed up by the nonprofit lab SRI International. They are seeking to track what forms of AI receive the most interest from researchers, as well as the number of engineers and dollars flowing into AI-based companies. Their goal is to release a comprehensive report on the state and rate of progress of AI development by the end of 2017.
Another, led by the Electronic Frontier Foundation (EFF), is compiling data from AI research to assemble an open source resource of progress. “We want to know what urgent and longer term policy implications there are of the real version of AI, as opposed to the speculative version that people get overexcited about,” says Peter Eckersley, EFF’s chief computer scientist, to Wired.
While some are still unsure what the value of this data will be, Eckersley believes that its use will become clear over time; he cited the example that tracking might lend real data to arguments over whether automation is taking away human jobs. In his data, he also already sees how vital it will be to track this technology as it advances. “The data we’ve collected supports the notion that the safety and security of AI systems is a relevant and perhaps even urgent field of research,” he says.
AI has the potential to be the single greatest human achievement, yet critics can also argue that it has the potential to be the most destructive. Having a clear-eyed view of the developmental process can give a certain level of assurance that AIs will not suddenly turn on us. However, there is no guarantee that this will be enough. While we are currently evaluating the need to place a speedometer on AI, as Wired put it, we may also need to consider if we need to be setting a speed limit.