预计阅读本页时间：-
ENVOI: A SHORT DISTANCE AHEAD—AND PLENTY TO BE DONE
Vice President of Engineering, Google DeepMind; cofounder, DeepMind Technologies
广告：个人专属 VPN，独立 IP，无限流量，多机房切换，还可以屏蔽广告和恶意软件，每月最低仅 5 美元
AI researcher; cofounder, DeepMind Technologies
Head of applied AI, Google DeepMind; cofounder, DeepMind Technologies
For years we’ve been making the case that artificial intelligence, and in particular the field of machine learning, is making rapid progress and is set to make a whole lot more progress. Along with this, we’ve been standing up for the idea that the safety and ethics of artificial intelligence is an important topic that all of us should be thinking about very seriously. The potential benefits of artificial intelligence will be vast, but like any powerful technology these benefits will depend on the technology being applied with care.
While some researchers have cheered us on since the start of DeepMind, others have been skeptical. However, in recent years the climate for ambitious artificial intelligence research has much improved, no doubt due to a string of stunning successes in the field. Not only have a number of longstanding challenges finally been met but there’s a growing sense among the community that the best is yet to come. We see this in our interactions with a wide range of researchers, and it can also be seen from the way in which media articles about artificial intelligence have changed in tone. If you hadn’t already noticed, the AI Winter is over and the AI Spring has begun.
As with many trends, some people are a little too optimistic about the rate of progress, going as far as predicting that a solution to human-level artificial intelligence might be just around the corner. It’s not. Furthermore, given the negative portrayals of futuristic artificial intelligence in Hollywood, it’s perhaps not surprising that doomsday images still appear with some frequency in the media. As Peter Norvig has aptly put it, “The narrative has changed. It has switched from, ‘Isn’t it terrible that AI is a failure?’ to ‘Isn’t it terrible that AI is a success?’”
The reality is not all that extreme. Yes, this is a wonderful time to be working in artificial intelligence, and like many others, we expect that this will continue for years to come. The world faces a set of increasingly complex, interdependent, and urgent challenges, requiring ever more sophisticated responses. We’d like to think that successful work in artificial intelligence can contribute by augmenting our collective capacity to extract meaningful insight from data and helping us to innovate new technologies and processes to address some of our toughest global challenges.
However, in order to realize this vision, many difficult technical issues remain to be solved, some of which are longstanding and well known in the field. While difficult, these problems can be overcome, but it will take a generation of talented researchers equipped with plentiful computational resources and inspired by insights from machine learning and systems neuroscience. While this is likely to disappoint the most optimistic observers, it will give this community some time to come to grips with the many subtle questions of safety and ethics that will arise. So let’s enjoy this new sense of optimism, but let’s not lose sight of how much hard work is left to do. As Alan Turing once said, “We can only see a short distance ahead, but we can see plenty there that needs to be done.”
