October 23, 2023

AI-First, Not AI-Later: How To Win In The Future Digital Economy

- The winners are likely to be those with the foresight to look at the world through the lens of cognitive automation and augmentation.

Artificial intelligence is taking the world by storm. Machines are now capable of solving many day-to-day cognitive and creative tasks that we believed would be reserved for humans brains for some time yet.

What the industrial and digital revolutions did for physical work, the currently ongoing cognitive revolution is doing for cognitive work. And since cognitive work is found absolutely everywhere, the implications of this are enormous and crucial to understand for any business that operates in the digital economy.

All paradigm shifts create winners and losers, with everything from large corporations to small startups claiming their share. For instance, the rise of a giant – Amazon – throughout the digitalization era led to the decline of traditional bookstores like Borders. Meanwhile, in travel planning, a large number of online platforms like Expedia and Kayak disrupted established brick-and-mortar agencies.

While Amazon’s is the story of a monolith, the travel industry’s transformation showcases the collective power of smaller digital disruptors. Both of these scenarios can be expected to happen in different sectors in the cognitive era as well. But what will separate the winners from the losers this time?

From Digitalization to Cognitive Automation and Augmentation

The winners of the cognitive revolution are likely to be those with the foresight to look at the world through the lens of cognitive automation and augmentation instead of simple, improved digitalization, and the adaptability to take attractive positions in value chains that promise to solve cognitive work for people using AI. This is an AI-first strategy.

The losers will be those that either myopically choose to remain in the digitalization era, or simply fail to break free from its shackles. The myopic group will have employed the AI-later – or even the AI-never – strategy, and the failed group will typically have employed an AI-on-top strategy where AI has been shoehorned into products of digitalization-era value chains. Both categories of losers will see their competitive advantages, margins and profits erode and even disappear; the former because they are playing the wrong games in the wrong league, and the latter because they are playing using the wrong formations, strategies and tactics.

As Wayne Gretzky famously said: “I skate to where the puck is going to be, not where it has been.” In the digital economy of the past few decades, the metaphorical puck –  the money – has long been in value chains organized around digitalization. This metaphor and its usage is borrowed from the book The Innovator’s Solution by Clayton M. Christensen and Michael E. Raynor, which is a seminal work in the area of disruptive innovation and inspired and helped contextualize the ideas in this article.

The concept of value chains is central both in their work and in economics in general, and there are many different definitions of a value chain that depend on what is the unit of analysis. For the purpose of this article, a value chain can be defined as the configuration of resources and activities required to solve a problem – both in B2C and B2B settings – and the companies that possess those resources and conduct those activities.

The value chains of the digitalization era broadly solve problems in categories such as providing access to information and entertainment, facilitating communication, collaboration and productivity, and providing digital access to physical goods. In the coming years, commoditization will claim most of the remaining digitalization-era value chains it has not yet claimed, and the puck will instead be in value chains organized around the automation and augmentation of cognitive work.

One could argue that it is possible – and even natural – to presume that AI is simply a way to solve all of the problems that the existing value chains already solve in a better way. While in some cases we may in fact end up with cognitive value chains that are simply improved versions of more or less identical digitalization-era value chains, not considering the potential for fundamentally different value chains when a transformative technology comes into play is a sure-fire way to miss a potentially open goal.

The signs of this are already there. We are witnessing the advent of foundational AI models and autonomous AI agents, and we can glimpse the presence of AI-optimized hardware on the horizon. We are also realizing that developing AI products is fundamentally different from developing non-AI products, and can therefore start to design and build true AI-first user experiences.

All these things considered, it seems likely that the resources and activities and their configuration in value chains organized to solve automation and augmentation of cognitive work will be different from digitalization-era value chains.

Some examples that build on this intuition could be how AI-first user experiences for retail can more or less eliminate the role of social media in direct sales, or how generative AI can fundamentally change how content production works in digital entertainment. These value chains will clearly look different than their equivalents today.

The Nature and Evolution of Cognitive Work

In order to understand the revenue potential of these new value chains and what problems they might be solving, we should look at the proliferation of cognitive work and how it manifests in people’s professional and personal lives. There is certainly enough of it to go around these days; in most developed countries, it is what the majority of us are paid to do. 60% of US workers are considered knowledge workers, and there are more than 1 billion knowledge workers in the world.

Knowledge workers typically have little in the way of physical labor in their workdays. We spend our time analyzing information, communicating and collaborating, planning, creating digital artifacts such as documents and code, making decisions, and thinking creatively. We solve problems using our human intelligence, more often than not using digital technology.

In our personal lives, we also no longer have to visit brick-and-mortar retailers, banks, travel agencies, and cinemas. Instead, we spend our time on digital tools and platforms wading through heaps of information in order to make informed decisions about what to buy, what to have for dinner, where to place our money, where to travel, and what to read, listen to and watch for entertainment, for learning or simply for keeping up with the world.

It is clear that both our professional and personal lives are filled with cognitive work. Some of this happens on digital interfaces of varying quality, and some of it happens completely outside of digital platforms. Some of the cognitive tasks are simply chores that most of us would like to get rid of, while some are in essence motivating – even meaningful – parts of our work and our lives. The cognitive value chains of the future will be organized around solving both of these types of tasks.

Strategizing for the Cognitive Era

This categorization provides actionable advice for companies that recognize that they already are – or soon will be – solving cognitive tasks for consumers, and wish to position themselves for the cognitive era.

  • For cognitive tasks which do not yet have good enough digital interfaces, or which do not have digital interfaces at all, the metaphorical puck will still remain in digitalization value chains for some time. This could be within traditional digital laggard industries like legal services, real estate, construction, and government services, education and healthcare in most countries. Companies can still take attractive positions here for some time without an AI-first strategy; and indeed, many are, and more will. However, companies that take an AI-first strategy even in this situation will be much better positioned for the inevitable time when digital interfaces are good enough and the puck is in the cognitive value chains, because their products have already been built with cognitive automation and augmentation in mind. In short: an AI-later strategy is viable in the short term, but an AI-first strategy is preferable in the short term, and necessary in the long term.

  • For cognitive tasks which have good enough digital interfaces and solutions, the metaphorical puck will be entirely in cognitive value chains. Currently, this is mostly within digitally mature industries like e-commerce, online advertising and marketing, digital entertainment, and to some degree financial services; at least in the customer-facing parts of their business. Competing on “better digitalization” is not very profitable, and companies can only take attractive and profitable positions using an AI-first strategy both in the short and long term.

Tesla is a prime example of a company that takes an AI-first strategy through attractive positions in both industrial and digitalization value chains, in order to be in the best possible position for when the puck inevitably will be in a cognitive value chain: self-driving.

In addition, we should consider the two types of cognitive tasks that the value chains will be solving.

This is likely to have less impact on the exact configuration of the value chains; a value chain for cognitive automation might look very similar to a value chain for cognitive augmentation. Still, it has major implications for what an AI-first strategy should be oriented towards, which in turn makes it clear which value chain a company will be competing in. This is most relevant for companies in the end-user facing part of value chains.

  • For cognitive tasks that are mostly considered chores, the value chains will likely be configured to enable cognitive automation, and the AI-first strategy must therefore be oriented towards cognitive automation. In this category, the best user experience is no user experience at all, and the optimal solution is for AI to in the end just solve the problem with no direct input from a user. This mostly has implications for how the product is designed, what data needs to be collected, which technologies should be chosen; it primarily informs the product and product development strategy.

  • For cognitive tasks that are motivating or meaningful in themselves, the cognitive value chains will likely be configured to enable cognitive augmentation, and the AI-first strategy must therefore be oriented towards cognitive augmentation. In this category, the best user experience is the one that makes the cognitive tasks feel maximally enjoyable and/or meaningful to the user. The user experience should enrich the tasks, not remove them. Like in the previous category, this primarily informs the product and product development strategy, but the product design, data and technologies will be different.

It’s important to note that it’s hard to predict exactly what the cognitive value chains will look like at this point, which means companies need to be cautious about being too deliberate with the details of their strategies.

History is filled with examples of companies that have burned themselves by tunnel-visioning towards the wrong landmarks in emerging competitive landscapes with unknown topographies. It might therefore seem like the advice to pursue an AI-first strategy at this point is premature, and even ill-posed.

However, the only assumption one needs to believe for this advice to make perfect sense is that solving cognitive tasks for people using AI will be the primary basis for competition in future digital products. This means that value chains will organize – or are already organizing – around this type of end-user value proposition, and that companies need to prepare to skate to where the puck will be.

Skating Ahead

Deciding on an AI-first strategy is like deciding on the right formation to play on the rink before a game. This needs to be deliberate in order for the team – the company – to be synchronized in their pursuit and defense of the puck, and it needs to be based on a reasonable hypothesis about what will be competitive.

However, the specifics of your strategy need to emerge and adapt from the matches you play and the teams you face. Companies don’t necessarily know what the value chains will look like, but if they know that they will be contributing in some way to the solving of cognitive tasks, their strategies need to be AI-first.

Even coming to this conclusion, it can be tempting to postpone decisions about how to prepare and position a company for the cognitive era. But in reality, not making a decision is also making a decision.

Even if the puck is currently in your possession – i.e. you are making good money in a digitalization value chain – you can be absolutely certain that it will be elsewhere on the rink at some point – i.e in a different part of the value chain, or in a different value chain altogether.

To chase where the puck has been and be consistently behind is a great way to tire yourself out; you may be disrupting your opponents, but you’re burning resources and you’re not winning. Not even trying to skate to where the puck will be? That’s a great way to get benched and forfeit the game.

In conclusion, the only strategy that is likely to win in the digital economy of the cognitive era is an AI-first strategy, because it recognizes that the game will eventually be played over cognitive automation and augmentation. Simply realizing this is a key milestone in a digital company’s journey, and it’s the first step towards staying relevant in these times of extraordinarily rapid technological progress.

AI-First, Not AI-Later: How To Win In The Future Digital Economy