The transformation of the computer | MIT Technology Review


Wing himself did not intend to study computer science. In the mid-1970s, he entered MIT to pursue electrical engineering, inspired by his father, a professor in that field. When he discovered his interest in computer science, he called her to ask if it was a passing trend. After all, there are no such books in the field. He assured her not to. Wing and major changed and never looked back.

Former corporate vice president of Microsoft Research and now executive vice president for research at Columbia University, Wing is a leader in promoting data science in many disciplines.

Anil Ananthaswamy Wing recently asked about his ambitious agenda to develop “reliable AI,” one of 10 research challenges he was recognized for his attempt to make AI systems more fair and less biased.

Q: Would you say that there will be a change in the way it is calculated?

A: True. Moore’s Law takes us a long way. We know we’re going to hit the ceiling for Moore’s Law, [so] Parallel computing has become prominent. But the turning point is cloud computing. Originally distributed file systems were a form of baby cloud computing, where your files were not local to your machine; they are somewhere else on the server. Cloud computing takes that and makes it even bigger, where the data isn’t close to you; the calculation is not close to you.

The next move is about the data. In the long run, we fix cycles, making things faster — processors, CPUs, GPUs, and many more parallel servers. We ignore the data part. Now we need to fix the data.

Q: That’s the domain of data science. How do you explain this? What are the challenges of using the data?

A: I have a very short definition. Data science is the study of extracting value from data.

You don’t just give me a bunch of raw data and I push a button and the amount comes out. It begins with collecting, processing, storing, managing, analyzing, and visualizing data, and then interpreting the results. I call this the data life cycle. Every step of that cycle is a lot of work.

Q: When you use big data, concerns about privacy, security, fairness, and discrimination often arise. How does one address these problems, especially in AI?

A: I have a new research agenda that I am promoting. I call it reliable AI, inspired by the decades of progress we’ve made in reliable computing. By reliability, we usually mean security, reliability, availability, privacy, and usability. Over the past two decades, we have made a lot of progress. We have formal procedures in place to ensure the accuracy of a piece of code; we have security protocols that increase the security of a particular system. And we have specific ideas of privacy formally.

Reliable AI raises the ante in two ways. All of a sudden, we’re talking about stability and accuracy-stability means that if you mess up the input, the output is less messy. And we’re talking about translation. These are things we never talk about when we talk about computing.



Source link

admin

Leave a Reply

Your email address will not be published. Required fields are marked *