Alexa, Did the Man Eat the Hamburger?
By Amy Willis
Are you ready for the robots?!?!?
This week, EconTalk host Russ Roberts returned to the topic of artificial intelligence (AI), welcoming computer scientist Melanie Mitchell to talk about her new book, Artificial Intelligence: A Guide for Thinking Humans. So what should thinking humans such as yourselves be worried about with regard to AI? Or should we just sit back and chat with Alexa?
The conversation begins with Mitchell providing an overview of the AI landscape today. Many of people’s worries, she says, do not account for a fundamental misunderstanding of what knowledge is. Just what is it machines know, and what can they learn? For Mitchell, working on AI has taught her how hard easy things can be, and how easy hard things can be… What did you take from this week’s episode?
1- Roberts muses about what it might be like to have a machine that could cure poverty. But it’s a wicked problem … To what extent are problems such as poverty simply knowledge problems? To what extent might AI someday be useful in solving such problems?
2- Mitchell describes one of the unique characteristics of humans as having the ability to predict the future in mundane situations. (Think of the discussion of electric cars stopping suddenly.) Why can’t AI succeed at similar tasks? What is it we humans do to learn how to put that sort of story together? (Think about “explainability,” of the story of the female soldier with the dog, or the story about a man with a hamburger.) As AI gets better at such tasks, will it merely have made AI better, or will it tell us more about our brains?
3- What are “deep neural networks,” and how do they illustrate the convergence of big data and computing power? What are the current capabilities of such networks? How do they compare to Amazon’s Mechanical Turk? What do you think of the latter? Is it more an indictment of the state of the world today, or is it super cool?
4- How does thinking about AI help you think about human beings? How much of our intelligence is “invisible” to us? How worried are you about the implications of AI for humanity? Explain.
5- What are the worries of people from within the field of AI? To what extent is the level of anxiety of people such as Hawking, Musk, Bostrom, etc. just wrong, or at least still too far away? How can we make sure AI is aligned with our values?