‘Gradually, then suddenly’ - Part 2

Midjourney / Antony Slumbers

Where’s your moat?

How are you planning on future proofing yourself against a world increasingly mediated through AI?

In Part 1 we asked:

1. Are you thinking how to redesign your business for an AI mediated world?

2. Have you checked to see which of your customers are?

But we should also be asking similar questions as individuals.

Because intrinsically we should, but also because each of us needs to assess how our answers differ, or not, from the same answers when asked about the companies we work for, or with.

After all, our personal incentives are somewhat to very likely to be different, and as the wise Charlie Munger used to say ‘show me the incentives and I’ll show you the outcomes’.

Let’s just take one example.

We often hear people say ‘’AI won’t take your job, but someone working with AI will’, but this is rather simplistic. What if, by working with AI:

1. We find our jobs commoditised? From various analyses it appears to be that Generative AI benefits below average workers more than it does top tier ones. So whilst that means a levelling up across the board, it also means a greater supply of people are now capable of doing X, augmented with AI, than previously. And when you have a rise in supply with no rise in demand the price for those services goes down. Simply put, employers won’t have to buy in top tier talent to do X anymore, so average renumeration is likely to decrease.

2. By working with AI, being augmented by it, one is also training the AI. Explaining in granular detail how A impacts B, or X leads to Y. Over time, as we know, AIs improve as they are fed more data to learn from. Whilst not currently a major force, the development of AI ‘Agents’ is rapidly developing and these ‘Agents’ are being designed to be combinatory. I.e they can be plugged together like Lego. Agent 1 does this, then hands over to Agent 2 to do that. So, whilst today we talk about AI only impacting tasks not jobs, this may well change when multiple tasks can be combined into one action, or AI workflow.

So in both these cases being augmented by AI is a net positive for employers, and a net negative for employees. See what I mean by incentives?

How then does one respond?

There are many possible ways, but there are only three core directions of travel:

1. A certain type of company will push ahead with augmentation hoping for exactly the results noted above, and hoping that their workforce does not click they are training their replacements, or actively reducing their value. I expect to see a lot of this. Sadly, certain types of job, comprising very large numbers of workers, such as commodity call centres, will ‘fit nicely’ into this management strategy. Much senior management of course will be highly incentivised personally to adopt this approach. Getting rid of employees is a great way to boost the bottom line - look at how effectively large tech companies have done this over the last two years.

2. A different type of company, with a different type of employee/employer relationship, especially around areas of ‘Trust’, will seek to pursue a third way of working with AI. This will be to lean in to commoditisation and substitution of their existing ways of working, but to do so with a view to redesigning these workflows to enable more and/or new value creation. With the underlying mindset being that with these tools we can create better/cheaper/faster. That, as in all previous technological ‘phase changes’, creative destruction has occurred and on balance, over time, society has benefited.

3. Or alternatively, even disregarding the development of much in the way of new products or services, employers and employees embrace all these trends and try to grab market share. For a time at least, until the market catches up, there is sure to be the opportunity for companies who are super productive to be super competitive. Most likely at the expense of the Type 1 companies mentioned above.

Ultimately, at an individual level, you need to concentrate on two things:

1. Understanding the incentives of the company or companies you work for, or with. What is the management ethos? How are they likely to embrace AI? Do you trust them with your labour? If not, then plan to move on. You’ll only be disappointed if you don’t.

2. And even more importantly, think hard about how and where your value resides. What can you do that the machines cannot? Which skills can you develop that will remain premium? Don’t kid yourself with ‘they’re just stochastic parrots’ or ‘they only produce average, generic content’. Often, average, generic content is enough and anyway these tools are improving exponentially. Assume they can do more than most think, and plan accordingly. Pay special attention to where you think value will either remain or move to - for example Uber commoditised knowledgeable taxis drivers, but greatly benefited the designers of Uber’s software and business. Remember Clayton Christensen’s "Law of Conservation of Attractive Profits." which posits that as industries evolve, particularly through processes of modularity and commoditisation, the locus of attractive profits tends to shift along the value chain. There’s always profits somewhere - that’s what you’re looking for.

Looking after number 1 needs to be your starting point. And if you are looking at all of this from the company point of view, look out for people who are looking after number 1. They’ll be best at helping you navigate to a prosperous future.

So much of the future will be determined at the regulatory level (look out especially for Laws levelling out tax treatment of Capital and Labour), and many things are already written in stone in terms of direction of travel, but as individuals we still do have great agency, and we must use it.