The Transformation of Knowledge and Power in the Digital Age

As we get better and better at training machines to emulate humans, are there certain aspects of human intelligence that artificial intelligence will never be able to copy?

Neil D. Lawrence is a professor of machine learning at the University of Cambridge. His new book, The Atomic Human: What Makes Us Unique in the Age of AI explores the meaning of intelligence as it relates to both humans and machines. 

Neil and Greg chat about the nuances of human intelligence and artificial intelligence, discussing how terminology affects perceptions and expectations of AI, pivotal technology advancements in history that paved the way for AI, and the insights Neil gained from his time at Amazon. 

*unSILOed Podcast is produced by University FM.*

Show Links:

Recommended Resources:

• Centrifugal governor
• Bletchley Park
• Tommy Flowers
• Joaquin Quiñonero Candela
• David A. Mindell
• Pierre-Simon Laplace
• Robert Solow
• Jeff Wilke 
• Geoffrey Hinton

Guest Profile:

• Faculty Profile at University of Cambridge
• Professional Website

His Work:

• The Atomic Human: What Makes Us Unique in the Age of AI
• Google Scholar page

Episode Quotes:

The trade-offs of increasing automation and the moral concerns of AI

25:16: As you increase automation, things that would have been moral judgments get moved into processes, whether that's courts of law or whatever; we tend to sort of codify what was a moral judgment, and it brings big advantages. It means we can live together at scale. It reduces the moral load we have if I can make a thousand employees redundant without having to worry individually about how many of them are single mums or whatever I'm worrying about. But, we lose something in that process. And one of the big concerns I have with AI is, yes, something like that's going to happen again. And I don't want to prejudge the future—what people will decide about where they want this technology automating decisions and where they want the human element in. But what I strongly feel is that, as a society, we're not being invited into that decision. And that decision is being made by very few companies and entities who themselves have proven themselves to have a very limited understanding of these subtle elements of society.

On the great AI fallacy

22:17: I think that the great AI fallacy was that we built anything that was going to adapt to us and accommodate us. When we hadn't, it was just more automation of things that humans had to do or could do in the past; but humans then had to accommodate this automation in order to make the best use of it.

Debunking the myth of AI as infallible, all-seeing, and dominating

31:38: One of the problems with the international conversation now is that it's conflating these two things. It's like the thing that appears intelligent is being intelligent through copying our own evolution, our cultural ideas, but then people are assuming that alongside that it has this characteristic of always getting things right, which is just not true because these shortcuts and heuristics it's using are our shortcuts and heuristics, which we know can fail in different circumstances.

What’s the role of software engineers in the emergence of AI?

55:09: So, this modern scribe is the software engineer in terms of the modern scribe, the person who can translate human ideas into things that can be on machines. So it's almost an advance in terms of the computer's powerful technology; it's actually an unpicking of the democratization of information technology.  Because as more and more of our understanding of the world is stored in machines, we're entering a world where it's harder for lawyers and accountants, etc., to access the machine. But this latest wave of technology offers the potential to put that right, because this latest wave makes it possible for a regular human to talk to a computer.