Pretending We Can Put the AI Genie Back in the Bottle Is Pure Foolishness

Commentary

The concerns about artificial intelligence are real. I have raised them myself.

Today’s most powerful systems can find hidden flaws in software and networks and turn them into working attacks far faster than was possible not long ago. Industry associations promise rock-solid security; so did Equifax and SolarWinds. Keeping dangerous tools out of malicious hands lasts only until one employee, one stolen password, or one sloppy setting lets a copy out.

That is why I have argued for real laws, not polite pledges. We need to treat a leak as something we plan for—not something we hope never happens.

I believe we should seriously consider a Geneva Convention rule that puts autonomous machines with the final say to kill in the same category as chemical and biological weapons. Putting human beings and hard-to-rewrite controls in the critical control path of energy, military, and other life-and-death systems, even if it makes them somewhat less efficient, is a must.

The COVID-19 pandemic showed us that when revenue is on the line, corporations and their trade groups cannot be counted on to police themselves. Granting AI developers immunity from lawsuits is a really bad idea.

Although no political system erases all AI risk, a system in which leaders answer, however imperfectly, to voters; in which companies and officials can still be sued, investigated, and voted out; and in which law still rests on God-given rights is vastly better than a godless order in which a small elite is the law.

Enemies of the United States and malicious non-state actors will use AI to attack. China, other states, and non-state actors will definitely use AI to gain dominance and advance their agenda. 

Countering those attacks will require our own cutting-edge AI. Just as shipping “dirty” factories to China that treats the environment as optional did not green the planet, so will allowing Beijing to take the lead in rolling out the infrastructure necessary to dominate AI not make the United States or the world safer.

So even as we write the laws and build the safeguards that keep this technology a net good for human beings, it is unwise to consider, even for a nanosecond, letting China take the lead.

Unfortunately, with help from China, anti-AI feeling has reached a dangerous pitch. Some of that feeling is earned, and some of it comes from those demagoguing the issue: inflating dangers and damages.

Those demagogues, with a whole lot of support from the legacy media, have downplayed and ignored the work underway to correct problems and imbalances, all in hopes of demonizing AI to win the midterms.

To be sure, the AI buildout has left real marks: higher electricity bills as giant data centers strain a crowded grid, strain on water supplies in areas with limited water resources, and the fear that a town will pay for someone else’s power or water. 

However, state, local, and federal officials are answering with laws, ordinances, and contracts that put the cost on the companies: special electric rates so data centers pay for the plants and lines they require, rules prohibiting the use of evaporative cooling in regions of scarce water, and permits tied to hiring and infrastructure the town actually gets.

Large AI companies are buying into these ideas: pay for your own power, pay for grid integration, don’t foist the buildout on families and other businesses, and, increasingly, reduce power costs for the communities where they build their facilities.

The panoply of initiatives, laws, agreements, etc., won’t immediately remedy all the issues, but they are definitely heading in the right direction. Sure, there will always be some downsides to AI, but the downsides of ceding AI dominance to China and other unsavory entities are an order of magnitude more harmful than the temporary issues being hyped.

Chinese Communist Party officials must be rubbing their hands in glee as many have adopted the scare tactics China has been pushing in thousands of fake social media accounts. 

Indeed, OpenAI recently released a report documenting the Chinese regime’s covert social media campaign against American AI growth. Operators and automated accounts pose as a parent, a student, a farmer, or some other figure, writing posts designed to whip up fears about AI, along with a link to a real article about increased power bills, water shortages, etc. Of course, many people, if not most, don’t take the time to read the story, so the exaggerations in the post often gain the credibility of whoever wrote the actual story.  

A China-linked comic portrays a family shouting “This is unfair!” at a tycoon with bags of cash over a power bill, while a U.S. senator recently stated, “Oligarchs want you to pay for these data centers with higher water and electric bills.”

This is unfortunate, as it lends credibility to Beijing’s disinformation campaign.

AI is here to stay, and pretending we can stuff it back into the bottle is foolishness that will have dire consequences. Defending against malicious use and against machines that slip out of human control requires advanced AI of our own. Falling behind puts American citizens at far greater risk than the temporary growing pains of building out AI infrastructure.

Do not let midterm scare talk hand China a lead we may not be able to overcome. Further, we cannot let a handful of giant companies scare politicians into creating laws and regulations that allow them to dominate AI while freezing out smaller AI players. Shielding Big AI from lawsuits while writing rules only they can afford will cripple the AI innovation necessary to protect against malicious AI.

Views expressed in this article are the opinions of the author and do not necessarily reflect the views of The Epoch Times.