Skynet Should Run the Bank of England
I think that Terminator 3 is the best in the franchise. I won’t go into why here (leave that for another post) and I imagine most people will disagree – then again I think that Phantom Menace is the best star wars so there we are. Anyway, I find it intriguing because it deals with the actual turning on of the Skynet system to control the US defence system. SPOILER WARNING: it did not go well.
Last week I had a meeting with some officials at the Bank of England and we briefly discussed the implications of Artificial Intelligence for monetary policy and the potential challenges and opportunities for central banks. Below is a rough outline of what was discussed.
Challenges
We know that AI has the potential to be incredibly disruptive to the economy. The worst case scenario is that the AI will kill us all. The second worst case is that it is so disruptive to the labour market that there is mass unemployment and wealth and power are concentrated in the hands of a tiny elite while the rest of us live in grinding poverty. This will lead to disillusionment with institutions, the liberal political system, and a breakdown in law and order leading ultimately to a collapse in capitalism.
These are the worst case scenarios. However, there are some more realistic challenges in the short term. We do know that AI likely will have an impact on the labour market. That could lead to increased unemployment and have implications for how the government finances public spending. All of this would create a challenge for the Bank of England as it wouldn’t exactly be clear if it is prepared for such a shock.
We also know that AI will create new products and services. How will these be impacted by monetary policy and how will monetary policy impact them? How should the Bank respond? Does it actually have the tools it needs?
How will AI influence decision making by firms and households? What is the appropriate response from policy makers?
As you can see there are still a lot of questions which makes for an interesting discussion but obviously concerning if we don’t yet have all the answers.
Opportunities
AI will bring lots of opportunities for policy makers. Take headline inflation (which includes food and energy). AI can help to create new products and production methods which can lead to increased food and energy security. Obviously central banks main focus is on core inflation, but the headline rate obviously still does matter so if that remains low and stable that will certainly make life easier for central bankers.
This breakthrough in technology could also apply to lots of other goods and services. If they can be produced more easily and in a way that is less susceptible to shocks, this again will keep inflation low and stable.
As pointed out in the ‘challenges’ section, we don’t have all the answers and policy makers probably don’t have the tools they need to adequately respond to AI. However, they can use AI to help them. AI could help them to develop new tools and strategies for dealing with shocks and communicating with households and markets.
On a related note, a recent paper demonstrated that AI can be used to predict future interest rate decisions by the Bank of England. The researchers used ChatGPT to analyse the speeches of MPC members and it accurately predicted how the MPC would vote. Obviously more research needs to be done but this certainly has important implications for policy makers. If such methods are adopted by firms, banks, and households they may end up making wiser financial decisions which would be a good thing.
Finally, AI can help to improve forecasting. I’ve written before why it’s important that institutions like the Bank get better at this and it’s encouraging that it has brought in Ben Bernanke to lead a review. It is also welcome news that the ECB and other central banks are starting to use AI and machine learning in its work. As AI continues to develop it will get even better at analysing large data sets and so will be able to produce much more reliable forecasts.
That’s pretty much what we discussed. However, over the past week I’ve been thinking about what the long term implications are. My conclusion is that it could be possible to hand over control of monetary policy to AI.
I’m always cautious when people say that computers should run an economy. We should absolutely not attempt to centrally plan the economy, whether that be through politicians, officials, or a supercomputer. This is a really interesting paper on ‘TechnoSocialism’ which goes into further detail as to why this is the case.
In short, Hayek is still right. The economy is constantly changing as households and firms make choices based on their needs and preferences. AI can process information much more quickly than humans, but it lacks the ability to discover and process knowledge in the same way that markets do. Prices and markets matter as they allow billions of households and millions of firms to coordinate decision making and gather information about which goods and services are in demand and how best to procure them.
The free market is still the best way to run an economy.
However, what I am talking about here is not the running of the economy, but rather the role of monetary policy by a central bank or government in that economy.
The Bank of England and other central banks have received fair criticism in allowing inflation to get out of control and then being too heavy handed in their response. However, it has done a good job on average at keeping inflation on target during its independence. We also know that a rules based system for monetary policy that is independent of government is much more effective than any of the alternatives.
Granting operational control to AI as opposed to a committee of humans who are tasked with processing so much information and who ultimately rely on their best judgement is likely to mean that the decisions made are the right ones. I’m not suggesting we do this anytime soon, but certainly something to think about.
Jobs for the bots
There has been talk among Labour that it would levy a tax on firms which replace human workers with AI. I think this would be a very stupid idea. I wrote about why for the Evening Standard on Wednesday. You can read it here.
Ultimately, while I do believe that AI will disrupt the labour market, it is far more likely to produce far more jobs than it destroys. Back in February 2019 I wrote a long piece on this with two very smart Stanford computer scientists and I think it still holds up really well today.
Why Gove is right
I disagree with Michael Gove on a lot of things. However, he is absolutely right in his ambition to turn Cambridge into the Silicon Valley of Europe. I wrote a piece this week for the Bennett Institute for Public Policy at Cambridge on why, if done properly, it would not only be a good thing for Cambridge, it is an absolute necessity for the entire country. You can read it here.
I set out my case for allowing Cambridge to grow on here back in the Summer. Cambridge really is a special place and it is too important to be preserved in aspic and turned into a museum town.
Schwarzenegger for America
Arnold Schwarzenegger was in London the other week to promote his new book. In an interview with the BBC he suggested that he would be a great US President. Obviously he can’t be as we’d need a Constitutional amendment for that.
It is an intriguing proposition. He took a principled stance against the Republicans both in and out of office and was keen to push for environmentally friendly policies. He is also the epitome of the American Dream. He is an immigrant who came from nothing and achieved so much. It would be nice to see the US start to embrace the benefits of immigration once more.
Also, he introduced Milton Friedman’s Free to Choose on the TV back in the 90s. You can watch it here. You’re welcome.
Anyway, he can’t be any worse than Trump or Biden.
Thank you
It’s now just over one year since I started Opportunity Lost. I started it as a place to write stuff which I didn’t think I could convince an editor to pay me for because they would either be too long, a collection of ideas, of limited appeal, or downright wacky. I was worried that nobody would read it and/or I’d lose interest.
This is my 69th post and so many of you have subscribed and read it each time. What is more, some of you have started to make a financial contribution (which you can do here: https://www.buymeacoffee.com/opportunitylost) and it really does mean a lot.
Thank you so much for reading.
Ben