An often quoted paper by Erik Brynjolfsson and Adam Saunders starts with this great line: “We see the influence of the information age everywhere, except in the GDP statistics.” They go on to note that Information technology has been static at 4% of GDP for the past 25 years. But the economics of artificial intelligence might change the game in the 25 years to come. Let’s see why.
First, continue with Brynjolfsson’s and Saunders’ explanation of why IT hasn’t grown as a proportion of GDP:
The answer isn’t about quantity, it’s about price. The bits that comprise today’s information goods are supplanting the atoms that formed yesterday’s encyclopedias, movie theaters, music CDs and newspapers. Online information may be updated every minute of the day and accessible almost anywhere in the world, but its price is usually radically lower than that of its physical counterpart, if there even is a price.
So Information Technology isn’t growing in the GDP statistics because everything it touches becomes cheap or free. It’s the economics of zero marginal costs as applied to digital goods, which I’veposted about before.Great for consumers and society overall, a disaster for producers. See how Information Technology (at 4% of GDP) fits into the bigger picture below (from Wikipeda).
To see how more capable artificial intelligence might change this static 4% of GDP trend, let’s look at health care as an example.
IBM’s Watson computer famously beat a human at Jeopardy in 2011. But this same technology is a natural fit in health care, where it could be used by nurses or relatively cheap administrative staff to do the diagnostic work of doctors. In fact, IBM has already won contracts to do just that. Diagnosis and treatment will be accomplished with administrative staff using computer/voice interaction, and costs will plummet. This technology, or a similar one from a competitor, will completely upend the health care industry. Doctors will still be needed, but only for difficult cases. A good precedent here is computer chess, where computers got better than humans, but the combination of computers and humans together is even better than either alone. With that said, for everyday health care we won’t need this beyond human level of expertise. Computers with an admin aide will be better than doctors alone, which should be fine. Day to day health care will become like fast food or going to Walmart.
Now we know how doctors will react to this threat to their pay and status. As a cartel. If cab companies can block car services like Uber by calling them unsafe and dangerous, and New Jersey can continue to make it illegal to pump your own gas because it’s “dangerous”, think how much more successful doctors will be at this game. They’ll induce fear of medical software by portraying it as illegal death machines intent on destroying the economy. And this will work, at least for a while. It will be impossible to use “doctor” software within existing health care companies due to conflicts of interest. So we’ll need brand new start-up organizations where customer service comes first and doctors are no longer king. This will take time, but once adoption starts the change could be quick. It takes a decade to train a new doctor. And training a computer is likely to take a decade as well, as we saw in last week’s post about the singularity. But once software is created, copying it costs nothing and happens instantly. Marginal cost zero software can upend an industry in a relatively short time.
Going back to the original point of this post, what we’ll be seeing soon is Information Technology invading the largest sectors of the economy such as health care, law, insurance, transportation and finance. When this happens IT will finally grow as a percent of GDP because it’s effect will be ubiquitous, not limited to niches like newspapers or entertainment. In turn this means the problem of the richest 1% dominating the world will get worse since they will own this technology. And software will continue down it’s current path of eating the world.