Human-Centered Artificial Intelligence

The episode contains a reasonable flow of ideas - self-efficacy as the design north star, the Amazon faceted-menu as an agent-free model, airbag incident-reporting as an AI governance analogy - but the core thesis (direct manipulation good, agents bad) is restated repeatedly with diminishing returns, and long passages are consumed by Shneiderman re-explaining terms rather than advancing new claims.

“a study a few months earlier in the year, uh, showed that while programmers thought they were saving about 20% of their time, actually it took 20% longer”

“2,500 lives are saved every year but by use of airbags...But in the early days, about a hundred children, babies and elders were killed by inadvertent inappropriate airbag deployments”

There are genuinely fresh re-framings - 'machine guessing' deflating ML hype, retreating from 'trustworthy' to 'safer and more reliable,' the zombie-idea metaphor for agents - but the foundational argument (direct manipulation vs. agents) is Shneiderman's 40-year position and is not substantially evolved here; it won't surprise anyone familiar with his work.

“if we instead of say machine learning, we say machine guessing, you know, it somehow deflates the notion”

“alignment is a fade phrase. It's not easily measurable and it's another part of the magic, adds confusion rather than clarity”

Shneiderman is a genuine pioneer - coined 'direct manipulation' in 1981, co-founded the HCAI Lab in 1983, Apple consultant for five years, Steve Jobs visited his lab - and speaks from decades of practitioner and research experience rather than as a media personality; he is highly relevant and credentialed for this exact topic.

“the keyboard on your phone uh, derives from the work that we did in the late 19 uh 80s. Steve Jobs visited our lab to see what we were up to. And I was a consultant for Apple for five years”

“there's a famous debate between myself and Paddy Moss of MIT Media Lab in 1997, uh, which is reported in the pages of the ACM Interactions magazine about direct manipulation versus interface agents”

The transcript is reasonably well-evidenced with named actors (Shawn McGregor's AI incident database, Nancy Leveson's book, Google Cummings's Tesla case), real rulings (Air Canada Supreme Court, $243M Tesla judgment), and numeric data (airbag lives saved/killed, programmer time-loss study); a few claims remain asserted without sourcing and some numbers are hedged with 'I think' or approximate dates.

“there were a case in Canada the past year in which Air Canada's chatbot offered a customer a reduced rate for a flight...it went to the Canadian Supreme Court which said very clearly, yes, you are responsible for it”

“a $243 million judgment against them”

The hosts occasionally generate good observations - the Amazon 'open vs. closed search space' contrast is insightful - but most questions are broad invitations ('what's your take on alignment?', 'what do you think is the solution?') and there is no meaningful pushback or challenge to Shneiderman's positions despite several contestable claims; the conversation reads as a friendly tribute rather than a productive inquiry.

“So do you think there's anything new coming now since like chat and Transformer models?”

“I thought it was quite interesting when you compared it to the Amazon sort of interface where you look for something like blue jeans and then what you presented with is really an interface that opens up the search space, it doesn't close it down”