AI adoption is patchy, focus on services and UK productivity (with Anna Edwards; Tom Mackenzie; Guy Johnson)

Asks how Will invests in AI given his hesitancy about the rate of adoption.

Anna Edwards

Will Hobbs

Bubbles are rarer than they appear; big stock surges often reflect a step-change in technology. Valuations suck in talent and capital. The frontier is still jagged. Currently in the picks and shovels phase (memory stocks), but the next phase is about the knowledge economy.

McKinsey study on mobile phones: they were out by 100x because they couldn't see which problems would melt away.

Asks if there is evidence that AI adoption is worth the cost and providing ROI.

Tom Mackenzie

Will Hobbs

Evidence is patchy. Generative AI struggles to migrate out of mundane tasks. But productivity statistics in the US are already picking up, predating generative AI. The trend is higher in the UK too if you use the right labor force statistics.

Machine learning and generative AI forbeau are already being deployed in widespread areas.

Asks if there is more upside in services that adopt AI or in the industrial/physical AI space.

Tom Mackenzie

Will Hobbs

Hunch is services. The cake is potentially biggest there. Manufacturing already has dark factories; services have most jobs and superpowers to confer. Marginal costs to serve come down, more people get access to legal and financial services.

Points out that costs are going up, not down, which is needed for wider adoption.

Guy Johnson

Will Hobbs

True, but we don't know which problems will melt away first. Talent and capital are being sucked into the space, so there's a good chance costs will come down. Diversified pitch is key.

McKinsey mobile phone forecast was out by 100x.

Notes that the market is buying expensive picks and shovels, and AI input costs are very expensive. Companies like Anthropic are losing money.

Guy Johnson

Will Hobbs

There is a big market delusion: five guys at a poker table all thinking they will win big. Some will be disappointed. Prices have to come down. Models are not doing anything much different from each other, so pricing power is limited.

Damodaran talks about this.

Asks if for that reason we don't need to worry as much about debt levels in European economies, including the UK, where the drama around gilts is 'hysterical'.

Anna Edwards

Will Hobbs

The drama has been hysterical. The UK had a lull in growth after the GFC, but the trend in productivity growth is higher than perceived using HMRC data. If you assume post-WWII trend growth, debt-to-GDP disappears in 50 years. The UK is a knowledge sector superpower.

Many worries about the UK debt pile are an extrapolation that the growth of the last decade is all we will ever have.

SignalTube