Goldman’s artificial unintelligence

Receive free Technology sector updates

We’ll send you a myFT Daily Digest email rounding up the latest Technology sector news every morning.

Artificial intelligence is a slippery thing, full of hype, semantic minefields and meaningless extrapolation. And then there’s artificial intelligence sell-side research.

A new report by Goldman Sachs, “The Magnitude and Timing of the AI Investment Cycle”, tries to put some loose-but-hard numbers on the rise of the robots.

In a decision that ought to surprise no one, GS thinks AI is gonna be a really big deal:

Such a productivity boom would be akin to those that followed previous milestone technologies like the electric motor and the personal computer. And while the timing and magnitude of such macro effects—with important implications for growth, inflation, and long-run interest rates—remain very uncertain and notoriously hard to predict, our previous analysis found that in most realistic scenarios the effects will be economically meaningful.

Both history and economic theory help inform how such effects could start to influence the economy and global financial markets. Facilitating a large-scale economic transformation entails significant upfront investment in physical, digital, and human capital to acquire and implement new technologies and transition business processes—investments that will likely show up before adoption and efficiency gains reach the point of driving large productivity gains.

“[N]otoriously hard to predict”, eh? Shame. But at least the short term should be fairly predictable. What’s the approach?

A simple extrapolation of recent trends in AI investment—using both changes in corporate mentions of AI and our equity analysts’ revenue growth projections for key AI-exposed business units—suggests that AI investment could grow rapidly in the next couple of years, approaching $100bn in the US and $200bn globally in today’s dollars by 2025 (Exhibit 4). Despite this extremely fast growth, the near-term GDP impact is likely to be fairly modest given that AI-related investment currently accounts for a very low share of US and global GDP.

If nothing else, we suppose this level of analysis reveals one industry that could start leaning on AI pretty soon.