Should the government or private sector determine which innovations flop or take off in the marketplace?

　　Most American businesses and consumers would argue the private sector should make this call. However, when it comes to new artificial intelligence-based technology, the Department of Justice appears to believe that the onus should be on the former. Government oversight over this innovative technology rests on an expansive interpretation of antitrust laws. Moreover, its actions could have negative rippling effects on nearly all economic sectors.

　　AI has become a popular way of ensuring maximal efficiency in the global economy.

　　In short, entrepreneurs have developed software algorithms that utilize economic data (both backlogged and current data) to provide their user bases with a better, more comprehensive understanding of how differing consumer trends, seasonal changes, breaking news, and other factors affect the demand for their services and products. These software programs use that information to provide their users with pricing adjustment recommendations that they are free to take or leave.

　　To say that this AI technology has taken off would be an understatement. Car rental companies, airlines, and hotels use it to ensure their prices match current marketplace trends. Hospitals and city and state governments use it to help quell congestion and long wait times. Farmers are even using technology tomonitor and manage field variability,maximize outputs, reduce costs, and improve sustainability, a practice known as precision farming.

　　However, the widespread utilization of algorithmic AI has the DOJ worried that businesses might begin using it for price-fixing, and it has begun throwing the antitrust books at many of these algorithmic software companies. Its actions have included but have not been limited to an October amicus brief filed against hotels and an August suit against one of landlords’ preferred algorithmic AI software. The Western District of Washington’s December 4 action against a different rent algorithmic AI firm has only added further fuel to the fire. 

　　This growing movement against algorithmic AI at the DOJ is reminiscent of when the government went after the Microsoft web browser at the start of the century, citing similar antitrust concerns. Ultimately, proposed governmental antitrust actions against Microsoft did not get anywhere, and history has proven that to be a good thing. Competition in this space has been robust, and businesses and consumers have managed to determine for themselves which web browser is worth their time. In a matter of years, the same will prove true concerning algorithmic AI companies. 

　　Antitrust law more or less comes down to a single question: is this practice fair or unfair? 

　　The default answer is that any given practice is fair. Congress knew that defining “unfair” business practices would be akin to defining “bad” food. Each list could theoretically be endless. And, whether an item belongs on the list is highly subjective. Just as there’s a stark difference between well-made food that some customers simply do not enjoy and “bad” food, there’s also a big gulf between business practices that consumers view with skepticism and those that are truly unfair. That’s why food critics have jobs and antitrust enforcers are assumed to apply expert knowledge of relevant business conditions.

　　As reductive as this analogy may be, it makes an important point: enforcement of antitrust laws should be based on the actual intent of antitrust law, not mere feelings. Yet that simple, significant lesson has been ignored in the recent DOJ debates about competitors using common algorithmic AI tools to help inform their respective prices. In essence, the critics of algorithmic AI are complaining about the steak without it having yet been placed on the grill, let alone tasted it, and without disclosing, that they are vegetarians.

　　Landlords and apartment complexes compete against each other for tenants. Yet, in its RealPage case, the DOJ argues that different economic actors using the same AI software programs represent “tacit collusion.” That is, even if these marketplace competitors have not signed agreements to price-fix together—and even if they are not communicating directly with one another—they are still indirectly acting in concert by using the same technology. 

　　To borrow the DOJ’s words from an amicus brief it recently filed in support of an appeal after a district court dismissed one algorithmic AI suit, An agreement among competitors to use certain pricing algorithms to generate default or starting-point prices is per se illegal even if there is no further agreement on final prices.

　　This nascent period of AI tool development and deployment should evolve free from the government trying to put its thumb on the scale.

　　A proper analysis under Section 1 of the Sherman Act—the basis of suits related to AI pricing tools—would likely bring an end to the DOJ’s series of filings on this topic. 

　　A plausible claim under Section 1 must allege the defendant was a party to a contract, combination, or conspiracy, and that the alleged conspirators imposed an unreasonable restraint on trade.

　　On the first requirement, similar, but independent action amongst the parties does not suffice. An actual and cognizable agreement must exist. Hallmarks of such an agreement include concerted action, a unity of purpose, and a conscious commitment to a common scheme.

　　The DOJ further contends that the mere act of companies sharing data with a common software company gives rise to collusion. It argues that deference to AI pricing tools can be even more effective than the smoke-filled rooms of the past.

　　Courts and scholars have shot down this line of thinking. What qualifies as an agreement cannot be watered down to the point of two or more parties reviewing similar information and contributing data to common projects that allow for a more robust understanding of market conditions. This logic, if accepted by the courts, would chill what has become ubiquitous and commonly accepted business behavior.

　　Companies have long shared information with third parties. How companies use that information to shape their business strategy is up to them alone as an independent business decision. It certainly does not constitute “concerted actions” with “a unity of purpose.”

　　Accordingly, enforcers have struggled to show how consideration of AI pricing recommendations constitutes an unreasonable restraint of trade.

　　Businesses have long shifted prices in response to common information and trends. To penalize a business for consulting a pricing recommendation from a third party seems to go well beyond the intent of the Sherman Act. That isn’t to say that in some cases businesses may indeed develop and defer to an app that explicitly suggests binding, coordinated pricing decisions. When and if such agreements are made and restraints are imposed, then enforcers should pursue a remedy on behalf of consumers with great haste. The current cases against the use of AI pricing tools, however, scream of algorithmic overreach.

　　Other frivolous cases will follow. Most will suffer from similar defects. The businesses attempting to operate more efficiently and price more effectively will evade liability. However, such litigation is far from costless. Each suit sends a signal to developers of AI tools, as well as their potential users, that they’d best lawyer up. That price of doing business will not dissuade all AI innovators from forging ahead with pro-competition tools. Some will opt to instead scrap their plans. That possibility should concern us all and inform the enforcement of antitrust laws in this turbulent period.

　　Antitrust law should not turn on vibes. It may concern some that businesses are learning more about their respective markets. Such investigatory efforts are not novel, though, and should not be subject to regulatory scrutiny until accumulated experience suggests the need for intervention. We’re not at that point. This nascent period of AI tool development and deployment should evolve free from the government trying to put its thumb on the scale. The alternative will slow AI innovation and potentially deprive both consumers and businesses alike of greater choice and lower costs.