This wiki is associated with Cognito Mentoring, an advising service for learners run by Jonah Sinick and Vipul Naik. The wiki is very much in beta, so you're likely to find many broken links and incomplete pages. Please be patient with us as we continue to improve our offerings.
Please connect with us to offer feedback on the wiki content.
Social value of working in artificial intelligence and machine learning
This page discusses the social value of machine learning | View other pages on the social value of particular activities
This page discusses the social value of artificial intelligence | View other pages on the social value of particular activities
The cluster of fields loosely related to artificial intelligence, and in particular the field of machine learning, are likely to become quite important over the next few decades. On this page, we try to estimate the social value of going into these fields using our general criteria for evaluating the social value of work.
- Here are some of the problems that are currently being tackled through machine learning, and that could arguably be tackled better using better learning algorithms: data mining, speech-to-text, machine translation, recognition of objects in images and videos, face recognition, medical diagnosis, web search, improving personalized recommendations. All of these have major real-world applications.
- As crude evidence in favor of the hypothesis that machine learning is important, note that machine learning techniques have been used quite successfully by companies such as Google and Facebook for tasks such as improving search or the news feed, and both companies have started labs and projects (Google Brain, Facebook deep learning/AI project) and hired top academics in the area to work full-time at the companies. And in 2013, there were loads of articles about machine learning as the hot new thing (see the footnotes for the Wikipedia page on Google Brain, for instance). Part of this is hype, but there's probably something going on. And with more and bigger data sets, it becomes more potentially valuable.
- Anecdotally, it seems some of the recent breakthroughs that led to improved machine learning weren't all that conceptually difficult (e.g., nowhere at the level of sophistication needed for a mathematical breakthrough) but generated huge social value. This suggests that until very recently, the field had a lot of low-hanging fruit. This is getting picked up rapidly, but it's likely to still have a lot more low-hanging fruit than other parts of computer science.
- Machine learning is one promising approach to AI. It's not the only approach, but it's probably one of the few approaches where partial progress is being made and there are economic incentives to continue making more partial progress. It's less speculative than other trajectories to AI.