Welcome to the Cato podcast. I'm Ryan Bourne, Cato's R. Evan Schaff Chair for the Public

Understanding of Economics. There's a fierce debate at the moment, in economic circles at least,

about how much artificial intelligence is already affecting the job

market and workers' productivity. There's been a bunch of microeconomic studies of individual firms,

industries that seems to show big effects already. Some customer support firms have seen

14% improvements in output per worker after rolling out new AI technologies. Consultancies,

some PR firms, have seen similar gains. But it's not clear yet whether this is passing through

to overall economic performance. A lot of the macroeconomic big picture numbers seem difficult

to reconcile with the idea we've seen an AI-fuelled productivity takeoff just yet.

Even though labour productivity growth, that's the amount that we produced per worker,

was pretty healthy last year for the past 12 months, 2.8%.

That stronger growth trend actually started in 2022, before many of the generative AI technologies really took off.

And looking across industries, it's not clear how far AI is responsible for this labour

productivity pickup anyway. Now, perhaps there's another way for us to measure how impactful

this new technology will be, and a more simple one. How often it plays a central role

in current affairs stories? And I'd humbly suggest that events this past week show that AI is

really starting to have a big impact. The federal government, of course, is now engaged in a legal

dispute of one firm, Anthropic, over the uses of AI for military purposes. One proposed bill in the New York legislature

seemingly wants to disable chatbots from giving advice in a range of licensed industries.

And new polling this week by NBC suggests that AI is deeply unpopular among the population,

with a minus 20% net favorability rating.

For context, only two entities polled worse in that survey.

...