AI to accomplish what?

Many people start realizing that AI is an inevitable disruptor of society. It will enable true personalization at scale across every area of your life, from health, to entertainment, relationships, work and education.

The question I ask myself is: What do we want AI to ultimately accomplish?

I met with around 100 representatives from the public sector a while ago and as a starting point I asked them two questions:

1) Is society getting better? Everyone said yes.

2) Do you know exactly what you look at when you answer yes to that? No-one knew.

This is a major challenge in society, that we haven’t agreed on in which direction to run. What do we want? How do we want to measure societal progress? It often still is about Gross Domestic Product (GDP). Is that what we want AI to optimize society for?

What’s positive is that we’re starting to see initiatives such as UN’s Human Development Index (HDI), Grant Thornton’s KPIs of the Future, the Social Progress Index, OECD’s Better Life initiative, ESGs etc. We are trying to get there, which is great.

The reason this is so important in relation to AI is that AI will get us wherever we want to go. Faster than we can imagine. Therefore, it’s of utmost importance that we have a conversation and alignment of where we want society to head and how we measure if it’s successful or not. In the story Alice in Wonderland when Alice is lost she asks the Cheshire Cat which way she should take? The cat answers that it depends on where she wants to go. Alice answers that she doesn’t know. The cat says that in that case it really doesn’t matter which way you take. As a society, we need to know which way we want to go, because we are Alice, and AI is the Cheshire Cat.

The overarching thing I believe we should strive for and optimize society for is Sustainable Wellbeing. Wellbeing, since if people are increasingly well, then we’re doing something right and vice versa. Sustainable, since we cannot spend everything we have in terms of resources at once. Wellbeing is quite well researched, so we could start working in that direction already today. Here is a breakdown of what it could look like at an overarching level that is not intended to be mutually exclusive, collectively exhaustive:

Just imagine for example that we tell the AI that fuels social media to optimize for sustainable wellbeing instead of advertising revenue. Or at least both.

Today AI is not yet optimizing large parts of society such as total healthcare, education or governance. That said, it has the potential to do so in a not too distant future. Because of that, it’s critical that we agree sooner rather than later on what we want to optimize society for.

If we look at the areas where AI is already leading a lot of the optimization work we see a lot of discrepancies between people who are better vs worse at setting goals they truly want to achieve. One example is the AI companies use to find new customers using for example advertising via Google, YouTube, Tiktok, Facebook and Instagram to name a few. For most B2C companies, these types of digital channels often play the biggest part of new customer acquisition, and it’s AI led already today. Now, ask any CFO what they optimize their customer acquisition for. Ask for example a retailer, a bank, an entertainment company or someone else. Most CFOs are not clear about this because of the silos in the company. Even a CFO caring deeply about the financial development of a company isn’t optimizing on profit. That’s how hard it is to keep up with technological development.

In the same way, AI will run faster than the regulatory environment in society, where we are currently quite far from agreeing on what to optimize society for. This will lead to sub-optimization at best in the short- to mid term.

We need to start the complex and hard work of agreeing on to optimize society for, otherwise AI may optimize it for something that humankind doesn’t agree with. There is a lot more to say about this space, and deciding what to optimize for alone won’t solve the challenge as we also need guardrails for what means are allowed to work towards that goal. That said, guardrails won’t help if the direction is wrong.

The cover picture for this article was generated by OpenAI’s DALLE as an answer to the prompt “An animated image of the society AI wants to create”. My question to you: Do you agree? If not, let’s decide what that picture should look like, and let’s do so together. My bet is on Sustainable Wellbeing.