The capabilities of artificial intelligence (AI) are a matter of heated debate. Can it generate new things? Can it produce scientific discoveries, novel hypotheses, or original ideas? The answer remains unsettled. There have been reports of successful applications in scientific discovery, but skeptics argue that current systems primarily recombine existing information. Nonetheless, this debate may be missing an important fact: AI is already changing how we access knowledge. The significance of this change becomes clearer through an analogy from economic history.
Monetary revolution in barter economies
Before money was invented, all exchange was conducted through barter. Markets existed and people eventually traded with each other, but transactions were highly inefficient. Finding someone who sold exactly what you needed and, at the same time, was looking for exactly what you were selling was difficult. Merchants provided relief by stocking different goods, but that did not remove the underlying friction entirely. Production and exchange were tightly coupled, and benefits of specialization were limited by the costs of finding trading partners.
Money uncoupled production from exchange. A farmer no longer needed to locate a carpenter who happened to want grain. Once exchange costs fell, markets expanded. Once markets expanded, specialization deepened. Once specialization deepened, productivity increased. The most important consequences of money were therefore indirect. It transformed the organization of economic activity itself.
We observe similar dynamics in how AI is transforming access to knowledge. To explain this, let us first define two concepts. By “information” we mean facts. By “knowledge” we mean the ability to interpret those facts and draw context-specific inferences from them.
AI transformation in the knowledge market
In modern (pre-AI) societies, the knowledge market has operated like a barter system. Experts produced, interpreted, and distributed knowledge. Their value derived not merely from possessing knowledge but also from making it accessible. Those who needed specific knowledge had to locate the expert who possessed it and then compensate that expert, monetarily or otherwise, to obtain knowledge in a form applicable to their particular problem. Think of hiring an attorney. You do not merely purchase information about the legal system. You purchase the application of legal knowledge to a specific problem.
AI is reducing the transaction costs of knowledge exchange much as money reduced the transaction costs of goods exchange. For the first time in human history, large numbers of people can obtain context-specific guidance without entering a relationship with the individuals who possess the underlying expertise. This is not the same thing as access to information. Google already made that easier. AI reduces search costs, translation costs, and scale constraints simultaneously. More importantly, it applies information to the user’s problem, thereby making knowledge accessible.
Predictions from the monetary analogy
The similarities between how AI is transforming knowledge exchange and how money revolutionized goods exchange are not only useful in better interpreting what is happening now, they also help anticipate what may come next. Here are five predictions from the history of monetary exchange:
First, the scale of the knowledge market will expand dramatically. The immediate effect of money was to boost exchange. Goods that previously could not be traded because matching costs were too high suddenly found buyers and sellers. Markets expanded to previously isolated areas. AI may have a similar effect on knowledge. Vast amounts of knowledge already exist, but much of it remains economically inaccessible because locating, interpreting, and applying it requires substantial time and expertise. By lowering these transaction costs, AI will expand the effective size of the knowledge market.
Second, with a larger market and lower transaction costs, further specialization in the production of information and knowledge will be feasible. Monetary exchange boosted productivity by allowing farmers and blacksmiths to focus more narrowly on what they did best. While knowledge will become cheaper to access and be consumed at a larger scale, the primary production of information by scientists, firms, governments, and individuals will likely operate with a finer division of intellectual labor. With better market efficiency, the scale of such production will increase.
Third, AI may not necessarily replace experts, but it will certainly change their job descriptions. While the production of new information will likely expand, other roles such as gatekeeping access to knowledge will erode. As the cost of accessing knowledge falls, the value of experts will shift from facilitating access toward producing new information and exercising judgment.
Fourth, AI will change the nature of scarcity. In a barter economy, matching a trade partner was the main challenge. In a monetary economy, trust became the main concern: metal money could be diluted and wealth stored in paper money could be inflated away. Similar trust, credibility, and reputation problems will likely emerge in the AI-assisted knowledge market. AI dramatically increases the supply of analyses, recommendations, reports, and explanations. Information and knowledge become abundant. Credibility does not. Experts will still be needed to evaluate, certify, and lend credibility to knowledge.
Fifth, new institutions will emerge. Once money found its footing and markets expanded, it was not long before new actors and institutions, such as banks, emerged to support trust and facilitate exchange. Knowledge markets will likely face a similar evolution. While it is difficult to anticipate the exact nature of this transformation, we will likely require more institutions that validate knowledge, certify expertise, and sustain trust. This will have profound effects on current institutions.
Implications for current knowledge institutions
Universities have traditionally performed three functions: producing knowledge, distributing knowledge, and validating knowledge. AI increasingly assists with the first function and directly challenges the second. The third function may become far more important. Academic systems currently reward high-volume production. Metrics on publications, citations, grant dollars, and conference presentations dominate evaluation. That emphasis made sense when the primary constraint was generating and disseminating knowledge. It becomes less compelling when knowledge production and distribution become dramatically cheaper.
Going forward, universities may increasingly resemble the trust institutions that emerged alongside monetary exchange. Their comparative advantage will lie less in transmitting knowledge than in validating it, which depends heavily on their ability to provide credible signals of quality. To perform this role, they will likely need to safeguard their credibility fiercely by prioritizing quality rather than quantity. As knowledge becomes easier to produce and distribute, credibility becomes the scarcest resource.
