Just yesterday, Smart Home speakers were infuriating us with their confused and stubborn responses to simple questions. These days, the ascent of something closer to real artificial intelligence with the likes of ChatGPT and Google Bard brings the AI arms race to new levels of promise (and redemption for Smart Speakers)—while also inspiring trepidation.
The trepidation stems from potential abuse of AI. That unease should stem not from AI as such, but from the federal government’s dogged moves to capture the deployment and trajectory of the technology. Joe Biden’s recent White House announcement of “voluntary commitments” from several AI leaders and tech giants offers a taste of the ill-advised regulation to come AI’s way if the whisperings of judicious natural intelligence go unheeded.
In Forbes recently I described the artificial problems we stand to create by anchoring into place the premeditated and unasked-for government-business “Blueprints” for artificial intelligence cropping up (these warnings were picked up in the Wall Street Journal also: “Washington threatens artificial intelligence”).
Federal and elite industry assessments and calls for regulation almost uniformly ignore the elephant in the room, which is the prevalence of government funding in AI. Military, intelligence, university grants, corporate subsidies and sweeping government-contracting make up a large proportion of AI research, deployment, and investment funding.
Government funding is not capitalism, so talking of “leaving AI free” and “letting the market decide” is incoherent in today’s setting.
The players calling the shots almost uniformly support public-private partnerships (PPPs) and new laws, and they uniformly support “regulation.” In reality, an end to government funding and steering should be the first step. Nobody wants to hear that, however. Fortunes are being made in federal government contracting, yet many in the pro-liberty community behave as if these pro-regulatory proclamations commitments have not been made and are not the central fact of the day, much like the spouse blindly ignoring an affair.
Cartelization or federal subsidies can be great for tech outfits staffed by mortal human beings with management horizons to think about. But the approach will block untold numbers of upstarts not party to the blue-ribbon collusions, thereby possibly costing trillions in distortions, diverted innovation and lost wealth over the coming decades.
A related problem I discussed in Forbes is the inclination of some to proclaim that no “Wild West” for AI exists, that the technology is already regulated by pre-existing “consumer protection” laws and the like, as if these are acceptable fallbacks.
One has to set aside revelations like the “Disinformation Governance Board” and “Twitter files” to imagine government will not itself be the primary abuser of AI’s breathtaking weaponry, surveillance, tracking and regulatory/remote-control aptitudes.
Decades of toil have gone into erecting the case for Administrative State rollback, to affirming that meddling by three- and four-letter regulators was destructive and conducive to rent-seeking even before AI arrived on the scene.
To now forget all that sweat and tears, and to proclaim that the likes of the botched Securities Exchange Commission, Consumer Financial Protection Bureau or even the Consumer Product Safety Commission “already regulate A.I.” in some futile effort to forestall legislation is ill-advised to say the least, legitimizing the Administrative State anew.
While calls for fairness and bias restraints may commendable, they cannot be taken seriously coming from today’s federal government whose abnormal definitions of these terms incline it toward curtailing free speech and the imposition of polices rooted in progressive biases. Politically motivated regulations that stifle dissenting viewpoints and hinder the genuine progress AI can offer are not a “maybe,” but intentional and underway.
Without the right war-gaming, not only will AI resume its absorption into Administrative State governance, prior work on corralling the insatiable regulatory state will be set back decades.
Biden’s initiatives ignore the vitality and necessity of competitive discipline. While proclaiming to pursue safety, security, and trust in AI systems, government domination typically means indemnification rather than safety, with top-down supervision stifling innovation in ways that do not lend itself to quantitative measurement, an affliction already pervasive in the regulatory state.
Our task is not avoiding the “overreach” in AI regulation we hear endlessly about, but recognizing that most of the rest of the Administrative State is overreach, and acting accordingly.
The proper A.I. program has one Big Bullet-Point Imperative, and that is banning of corporate and university subsidies. Any liberty focused agenda must start there. And as far as “transparency” is concerned, that should focus on misuse of data by government agencies, and the disclosure of AI funding sources and contractual obligations on the part of “private” businesses.
Until then, beware the seductive D.C.-speak of “striking a balance” between risk and innovation, The progenitor of misuse and abuse is more likely the phenomenon of government funding and the indemnification from liability that can accompany it. Potentially risky or dangerous technologies require market and competitive disciplines that the Administrative State waves aside for the benefit of rent-seekers and co-conspirators.
The oft-stated appeal to avoid a “top-down regulatory approach” is illusory in that top-down is inherent in the way we’re doing things now. The expansion of government control in AI is only one part of a far broader escalation of all-encompassing legislation covering infrastructure, energy, tech and more that are consolidating economic and social power in Washington. Much of this intervention has been predatory and abusive, undertaken during and in the wake of the COVID episode—and still ongoing.
The more prudent approach on all of these fronts is to curtail the prevalence of corporate subsidies, public-private partnerships, and government contracting, and to devise disciplines for legislators and would-be regulators whose actions are removing AI, crisis response, and much else from vital competitive disciplines.
The Administrative State’s tendency is to disrupt rather than discipline, worsened when government-corporate cartelization distorts the landscape. Without addressing federally inspired collusion and government funding of technology, libertarian-ish governance of artificial intelligence stands to be sidelined in favor of tech-dreck approaches and other forms of “good government” that translate into more government. The window of opportunity for a non-artificially intelligent departure from expansive government control will not last long.