An insider’s view on AI regulation

Recent Senate hearings on the need to regulate artificial intelligence (AI), along with some tech executives’ assertions that this industry should be regulated, have contributed to a tsunami of commentaries, proposals, speeches, emergency meetings and private conversations about the regulation of AI.

As someone who has spent most of his career at the intersection of tech business strategy and regulation, I’d offer some observations on what’s likely going on behind the scenes that will help make sense out of what will be a constantly shifting landscape of public positions on this important question.

It’s probably useful to recall that just as AI is emerging today, a similar situation arose in the mid-1990s, when the internet was emerging as a major alternative to conventional media as well as to private data networks like Prodigy. The question facing the tech industry then, as today, was how to regulate it, given the possibility that this new technology could be a very big deal.

It’s important to understand that both AI end-user services today and internet end-user services from the ’90s rely on a tech industrial supply chain that enables them. Component parts manufacturers depend on raw materials and supply equipment manufacturers who in turn supply service providers who then provide their capabilities to end-user AI service providers. Software creators depend on more basic software and then supply service providers, who then provide capabilities to end-user AI service providers. And end-user AI service providers rely on electronic delivery systems (public or private data networks), who in turn rely on underlying telecommunications services, equipment and software to enable them to deliver the end-user AI service.

Internet platforms and AI end-user services thus have parallel supply chains made up of thousands of separate tech businesses that make them possible.

Management of each business in these varied supply chains must take a look at the emerging technology (AI) and decide whether regulation might benefit their own business, and if so, who would be subject to regulation and what type of regulation. Consequently, over a period of years, there will be a confusing scramble of thousands of private evaluations along the lines of “What’s the impact of AI regulation on my business?”

Let’s recognize that many industry executives drive their business strategy and government policies by what they believe is best for humanity. These range from ideological views like “we always oppose any government regulation” to Google’s pre-2018 motto “don’t be evil.” Basing a government relations and business strategy on ideology rather than on business goals (like revenues, expenses, market share and net income) may be more tolerated in tech than elsewhere partly because some tech companies are dominated by individuals who don’t closely report to shareholders and can (for the most part) do as they please. Tech is also uniquely populated by a few enterprises that either are or are owned by nonprofit organizations. 

Finally, it is important to understand that what is financially best for any business’ shareholders over the short term may be different than what is in their financial interest over the long term; a lot depends on any tech executive’s planning horizon. Few businesses have a planning horizon of less than one year or more than seven years. However, although there are important exceptions, most tech executives carefully calculate what is best for their shareholders’ financial interests over the medium term and pursue that path in both business and government relations. The expression of that view is most often done through support for trade associations, think tanks, academics and activists who advance it — not public corporate pronouncements. 

Although most tech executives prefer an unregulated environment for the freedom it offers, perhaps surprisingly, some tech businesses conclude that it would probably be helpful if a supplier to them or a customer of theirs is regulated so they have extra leverage over the prices and terms of their own purchases or sales. Importantly, some tech businesses may conclude that their own regulation will benefit them because it will create a barrier to entry from new competitors. 

More generally, one’s assessment of such realistic scenarios as “no federal regulation” vs. “light federal regulation” vs. “thorough federal regulation” creates an entirely different analysis. “No federal regulation” can invite widespread state-level regulation and it can create pent-up end-user or media complaints leading more rapidly to “thorough federal regulation.” Alternatively, “light federal regulation” can sometimes pre-empt state regulation, create a framework for light international regulation and stave off “thorough federal regulation.” Finally, “thorough federal regulation” can sometimes lock in an acceptable regulatory framework in the U.S. and abroad for decades, closing the debate over regulations for a long time.

It now seems likely, but by no means certain, that end-user AI will have a major impact on the global economy and social structure, and will both create and destroy numerous businesses. In the process, there will be businesses that win and those that lose. Government regulation is not the key driver in determining business winners and losers, but it is an important factor. Depending on what is or is not done, it could easily be a key factor. Under these circumstances, we can expect a constantly shifting landscape of positions on whether AI should be regulated and if so, exactly who should be regulated and to what extent. 

Buckle up — these rides come only once every few decades.

Roger Cochetti is an award-winning executive in the technology and commercial space industries and a former U.S. government official. He served as a senior executive with COMSAT, IBM, VeriSign and CompTIA, has helped found a number of nonprofits in the tech sector, and is the author of textbooks on the history of satellite communications.