The Dual Use of AI – a Blessing for Some, a Nightmare for Others

A few years ago, I had a serious bicycle accident – broken bones, hospital, out of action for several weeks. Lying in my sickbed, I had time to think and wondered: Why was the operation so complicated, and why does it take so long to heal? Suddenly, I couldn't help laughing at myself: What a stupid idea! Why improve surgery and treatment? Wouldn’t it be better to avoid this whole damned accident in the first place?  

This thought has more to do with the effects of artificial intelligence (AI) than it might seem at first glance. It’s the eternal question of how to approach a problem: Find a quick fix, or better yet, eradicate the problem altogether.  

AI changes the world already

AI became widely available only in the last few years, rapidly gaining popularity by effectively solving numerous problems. But researchers are already exploring AI's long-term potential to address persistent challenges. Daron Acemoglu (Acemoglu, D. (2025). The simple macroeconomics of AI. Economic Policy, 40(121), 13-58.) shows how AI can increase productivity. This is typical for economists – aiming for lower costs, higher output, and greater competitive advantages here and now.

Of course, this benefits the customer. But the greatest benefit for the customer would be to avoid a problem altogether – to prevent the accident. The catch is: If consumers avoid the problem, companies lose the sale and revenue of the product. That’s not exactly business-friendly. 

Lessons from history

But the Kodak case shows that, sooner or later, things move in this direction anyway. Digital photography completely replaced the firm’s film business – billions of dollars in sales and company value vanished. Clayton Christensen’s called this The Innovator’s Dilemma: a market leader developed a significant innovation, but launching it risks cannibalizing its existing business. Apple handled this transition from iPod to iPhone smoothly: They simply killed the iPod themselves.

How will AI change this playing field? Technological breakthroughs are emergent and hard to predict. Schumpeter’s concept of “creative destruction” can only describe the phenomenon, but not help to manage it. This is unfortunate at the current stage, since AI could dramatically accelerate creative destructions. We must find a way to address this challenge.

The Trillion Dollar question

Could AI eliminate the causes of a problem and prevent the bicycle accident? For this specific case, we can imagine a solution – warning sensors or other technologies could help. But we need more insights how exactly can AI help on the path of avoiding such problems. This would change our world forever and ever. 

Our approach to the question

The logic is somewhat hidden. We need to look at consumption and production in context. Current economic models don’t cover this interplay between companies and private households. A new economic approach can help: nanoeconomics. It zooms in on the economic decision processes of single agents like consumers and producers. On one side, we have customers as consumers with a problem; on the other, companies as producers offering a solution. Although they are very different, they fundamentally operate by the same process logic: They use information, make decisions, and carry out activities, following the IDEA scheme (Information-DEcision-Activity).  

Since AI tools ultimately provide information, they come into play at the very first step. Ideally, AI-powered tools deliver better information faster and cheaper. (We’ll ignore the fact for now that AI is still in kindergarden and hence unreliable.) With better information the decisions will improve, the activities become more effective, and the results get better too. This improvement applies to both sides – consumers and producers – though not necessarily within the same transaction.  

This is where nanoeconomics comes in. It operates one level below microeconomics and examines the role of information in economic decision-making processes, as well as the effects of improved information supply. The core idea of the nanoeconomic framework is that better information leads to better decisions and activities, thereby avoiding mistakes and consequential needs. This saves costs and other resources, aligning with economic logic.  

A practical example – with scientific proof

Let’s take a practical example, an annoyance similar to a bicycle accident: cleaning an oven. The grease baked on over months is rock-hard – we need harsh chemicals and tools like scrapers and metal brushes. The work is exhausting, the smell awful, and it’s unhealthy too. However, there’s already a fascinating solution: ovens with pyrolysis. At 500 degrees, even the last bit of petrified grease crumbles and can be easily swept out. That sounds like a good solution, but it’s expensive, energy-intensive, and merely shifts the problem – it’s more clever, but just another problem solution.  

Problem avoidance would mean the oven’s interior doesn’t absorb grease at all – no adhesion. The oven stays clean forever. A dream comes true. How can AI help? It can support materials research. Aidan Toner-Rodgers (Toner-Rodgers, A. (2024). Artificial intelligence, scientific discovery, and product innovation. arXiv preprint arXiv:2412.17866.), an MIT doctoral student, shows in his research that AI significantly boosts innovation performance: In an experiment with over 1,000 scientists, the use of AI led to 44% more material discoveries, 39% more patents, and 17% more product innovations. The new materials were chemically more novel and radical – exactly what a grease-repellent oven material needs. However, AI mainly benefits top researchers: The top 33% of researchers doubled their output, while the bottom 33% barely saw gains. 

How AI can help solving problems

So we have it in black and white that AI can help produce better materials and hence can ultimately eliminate causes. This logic seems universal. We find “oven problems” everywhere – their causes lie in the material, construction, and functionality. Improve them, and the problem goes away. Of course, this requires expensive research, which must pay off. So the first products with these new materials are likely to be costly. But that’s another task for AI: finding cheaper technologies to produce these novel materials and products.  

This second use of AI is good news for consumers. Gradually, AI will help avoid many problems. But is it also good news for manufacturers? It’s as welcome as the news from Kodak’s development department to its top management that customers might switch to digital photos.  

The danger for producers

The danger for companies is concrete and real. If the oven stays clean, consumers no longer need cleaning agents. Sales of cleaning product manufacturers drop to zero. Logisticians stop transporting them, supermarkets stop listing them, and accountants at manufacturers, logisticians, and supermarkets have less to record.  

Will this happen for sure? The pressure comes from the customer, who no longer wants to clean. They pass this wish on to the manufacturer, who seeks solutions – pyrolysis or new materials. The pressure moves to suppliers, who develop better materials, and further to their suppliers. Everyone in the supply chain will likely use AI to find quick, cost-effective solutions. Whether this succeeds depends on many factors, especially market pressure. One factor could be that AI-driven job losses shrink household budgets, further motivating consumers to avoid expenses for solving problems like cleaning something. AI-powered shopping assistants will easily find the cheaper or problem-eliminating solutions.  

Time frame

This chain of effects is long and can take years. How often do we buy a new oven? But that doesn’t matter, as long as things move in the right direction: getting rid of the nuisances. This is the true dual use of AI: It helps solve problems (better healing methods) and, sooner or later, avoid problems (no accidents, no more oven cleaning). It’s hard to estimate how long it will take for this second track to take hold across the economy.

Until then, we have little choice but to ride our bicycles more cautiously.