More Choice Is Not the Same as More Wisdom

AI is quickly increasing what we can do. In a conversation with Bruce Lloyd, I began to wonder whether the harder task is deciding what is worth doing — and how much choice we really have.

By Bruce Lloyd & Frank Peters

I discovered something in Lyon almost by accident. I was walking through a part of the city I did not know, saw something that caught my attention and, instead of searching Google, opened ChatGPT. I took a photograph, asked what I was looking at and received an explanation. Then I did it again. Within a couple of hours, I had much more than a collection of holiday photographs. I had pictures, bits of history, my own comments and ideas for things I could explore later.

When I described this to Bruce, he immediately saw the possibilities. I could almost have created a small travel book during a two-hour walk. Then he asked a much more important question.

What was I going to do with all this?

That question quickly took us beyond travel. AI is giving us many more things that we can do. Bruce’s point was that, as technology gives us more possibilities, we may need to become more careful about deciding why we want to use them. Later he added another thought. Humans may still make choices, he said, but only between the choices they actually have the power to make.

I think those two ideas belong together. AI gives us more possibilities, but that does not remove the need for judgement, purpose or control. In fact, it may make those things more important.

When producing becomes the easy part.

A lot of the public discussion about AI is still about what it can do. Can it write an article, analyse a document, identify something in a photograph, translate a conversation, create an image or summarise a report?

More and more, the answer is yes.

But when these abilities become normal, producing something is no longer the most interesting part. If everybody can create ten drafts in the time it once took to create one, the real question is which draft is worth keeping. If AI can give me twenty ways to approach a problem, somebody still has to decide whether the problem itself is worth solving.

The same was true of my Lyon experiment. AI could give me information about almost everything I photographed, but it could not decide what was important, what I should explore further or whether there was a good story in it.

Bruce has been looking at the same issue in another way. He has been asking different AI systems the same questions and comparing their answers. He has found important differences, not only between Western and Chinese systems, but also between individual systems within those groups.

That matters because it shows that AI does not simply give us the answer. Different systems may understand the same question in different ways. Sometimes those differences are useful because they force us to think again.

This also changes the problem we face. In the past, producing and storing information took time and effort. That forced us to make choices. Today we can create and save huge amounts of information very easily.

My parents took slides. Years later, many of my mother’s slides had lost their colour and some had to be thrown away. Today we have solved that physical problem. Last year I came back from Korea with more than 1,500 photographs.

I have looked at them once or twice.

AI could sort them, describe them, add captions and probably turn them into a book very quickly. Bruce used to make photo books from his travels and believes he could now produce something much more detailed in a small part of the time.

But that still leaves the same question.

Would anybody look at it? Would I?

When producing things becomes easier, choosing becomes harder. We have to decide what is worth keeping, developing or ignoring. The danger is not only that we have too much information. It is that we start to believe that because something can be produced, it must have value.

Old problems inside new systems.

This is why I am careful when people speak as if AI has created completely new human problems. Many of the problems underneath it are very old.

I resisted buying a smartphone for quite a long time. People kept telling me what I could do with one, but I could already do most of those things on my laptop. Eventually somebody gave me the answer that convinced me.

“You don’t actually need it. It just makes a lot of things easier.”

That was enough.

The pattern is familiar. First a technology is unnecessary. Then it becomes useful. After that it becomes normal. Finally, we find it difficult to live without it.

The problems appear when a system works well for normal situations but cannot deal with something unusual.

Bruce made this point when we talked about automated systems. They can work very well inside the limits that somebody has designed for them. Then a problem appears that does not fit. A receipt is refused even though it seems correct. A journey does not follow the expected route. A customer has a reasonable problem, but there is no suitable choice in the menu.

At that moment, you need a human being who can understand what has happened and make a judgement.

Unfortunately, the human being is often the part of the system that has been removed.

This is not really a new AI problem. Organisations have always had difficulties with rules, processes and unusual situations. Technology can improve these systems, but it can also make them less flexible. If nobody is allowed to make a decision outside the process, then an efficient system can quickly become a frustrating one.

The same is true when organisations introduce change. Bruce described a familiar pattern. You have a system that works. You believe a new system could be better. But before you reach that better situation, things may become worse.

People know the old way. Then the process changes. Their normal routines no longer work and they have to learn something new. For a while, they become slower and less confident.

The learning happens during that difficult period.

Many organisations are not good at accepting it. People become frustrated, the new technology is blamed, support disappears and management changes direction.

AI makes this harder because the technology itself continues to change while we are learning it. A way of working that is successful today may need to change again next month.

One important skill in the AI age may therefore be quite simple: accepting that we will sometimes become worse at something before we become better.

The skills behind the technology.

This led Bruce and me to the question of jobs.

We hear a great deal about AI destroying jobs, and some kinds of work will certainly disappear. But new types of work will also appear, and some may arrive before we know what to call them.

I suggested that companies may eventually need somebody whose job is basically this: keep up with what is changing, decide what is important and explain it to everybody else.

Bruce immediately focused on the skills such a person would need. Technical knowledge alone would not be enough. The person would also need to understand how a new development affects the wider organisation and explain it in language other people can understand.

These are not new skills.

Curiosity. Judgement. Communication. Understanding the wider picture. Connecting ideas from different areas.

They also lead to another old problem.

Knowledge is power.

For many years, companies have treated knowledge management mainly as a technical problem. Build a system. Collect the information. Ask everybody to share what they know.

Bruce’s question is very simple. If my position in an organisation partly depends on knowing something that you do not know, why would I want to give that knowledge away?

A technical system cannot solve a human problem if it ignores the reason behind it.

AI may make knowledge easier to access, but it does not remove personal interests or competition. In fact, it may make them clearer. If information becomes easier to obtain, perhaps the value moves towards other things: experience, judgement, understanding consequences and explaining what really matters.

This is why “reskilling” cannot simply mean teaching people how to use the latest piece of software. Software changes. The deeper skill is learning how to keep learning while the technology continues to move.

More access, but who has the power?

There is a larger version of the same problem.

People often say that AI will make information more democratic because more people will have access to knowledge and powerful tools. There is some truth in that. A person with a phone can already do things that only a few years ago would have needed expensive software, experts or many hours of work.

But access and control are not the same thing.

Bruce’s point about choice becomes important again. Humans can only choose between the choices available to them.

Who owns the systems? Which companies will survive? Who decides how the models work? What happens if, after today’s competition, only a small number of large AI systems remain?

At the moment there is real choice. Bruce’s comparisons show that different systems can give different answers and approach the same issue in different ways. That allows us to compare, question and disagree.

But markets often become smaller over time as strong companies grow and weaker ones disappear. We could end up in a world where billions of people have amazing access to information through a small number of systems controlled by very powerful organisations.

More access for users, but more control at the top.

That does not mean AI is bad. It simply means we should be careful when we talk about it making knowledge more democratic.

Technology may reduce some differences in access while leaving differences in power untouched. It may even create new ones.

I suspect we are only beginning to understand what that could mean.

Thinking about thinking.

Towards the end of our conversation, Bruce and I somehow arrived at Marx, Engels, politics, wealth and artificial intelligence.

I suggested a question. Bruce changed the wording. Then I noticed something interesting.

Before we asked AI anything, two humans had spent time deciding what the question should actually be.

I thought that was important.

Bruce agreed, but added another point. Once the question is asked, different AI systems may understand it differently. Those different answers can move the discussion in directions we did not expect.

That is one of the things I find most interesting about AI. Used well, it does not only answer questions. It can also make us question our own thinking. It can show us an assumption, give us another view or make us disagree with it.

But there is also a risk.

Bruce often thinks about the difference between data, information, knowledge and wisdom. He has noticed that because ChatGPT knows he is interested in this idea, it sometimes brings the same framework into subjects where it may not really belong.

In other words, AI can challenge our habits, but it can also learn them and give them back to us.

That is an important difference.

The human task is not only to write better prompts. We also have to notice when AI is helping us think and when it is simply becoming very good at giving us the kind of answer we already like.

The conversation with Bruce was useful because he did not simply agree with me. He kept making the questions more difficult. More choice does not always mean more control. More access does not remove power. Better technology still has to deal with human behaviour, organisations and the uncomfortable period when learning makes us less efficient before we improve.

And greater ability does not remove the need for judgement.

That seems like a useful test for AI.

Does it simply help us produce more, or does it help us ask better questions, see our assumptions and make better decisions about what deserves our attention?

AI will almost certainly become more powerful. Bruce’s view is that humans will need wisdom if we want to use that power well.

I would be careful about claiming that humans have always shown great wisdom.

But more choice without good judgement is simply more noise.

And however clever the technology becomes, we still have to decide what is worth listening to.