AI Brinkmanship

A prophetic warning?

H+ 2004

Hiking alone in the High Sierras a few years ago, I was daydreaming while walking somewhat nonchalantly near a mountain top at 11,000 feet. Suddenly I reeled back in shock and terror. With one more step I would have fallen over a cliff into a chasm, a vast emptiness that went straight down at least 3,000 feet. With my heart beating faster than an airplane propeller, I stood aghast on the precipice.  I had reached the brink and, Gott sei dank, I had stopped at the brink. Whew!

Geoffrey Hinton, nicknamed “Godfather of AI”, has similarly stopped at the brink. Hinton pioneered understanding neural networks and shaped artificial intelligence systems powering many of today’s products. He just announced his resignation from Google. Hinton no longer plays AI brinkmanship. Why? Because generative AI is “scary.”

The Arrival of ChatGPT and OpenAI’s announcement of GPT4 has placed the world’s techies on the AI brink. The voices of Elon Musk, Steve Wozniak, and a chorus of in-the-know computerettes are chanting: pause AI development for six months while we think this over. Here’s their open letter: Pause Giant AI Experiments: An Open Letter. I added my signature to this letter.

Brinkmanship is the practice of following a dangerous trail toward the limits of safety before stopping. Stop playing AI brinkmanship! That’s today’s prophetic message.

I foresee two brinks, not just one. The first brink is the Singularity–that is, an AI takeover of planetary communication and decision-making. The second brink is more immediate–namely, malicious use of AI power to pillage and destroy.

AI Brink #1: The Singularity and the Takeover by Superintelligence

Our transhumanist friends invite AI brinkmanship, because they envision utopia just beyond the brink. If by the year 2045 superintelligence takes over, as Ray Kurzweil has forecasted, humanity will cross the Singularity threshold when technological growth becomes uncontrollable and irreversible. A new posthuman species will be born and take over management of our world’s systems. Singularity will open the door for a future techno-utopia.

But those who find AI brinkmanship “scary” such as Hinton are issuing a warning. Let’s keep AI management in human hands! To prevent AI brinkmanship, we need guardrails put up by ethics and public policy. Here’s how the Open Letter assesses our situation this side of the brink.

“Contemporary AI systems are now becoming human-competitive at general tasks, and we must ask ourselves: Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization? Such decisions must not be delegated to unelected tech leaders. Powerful AI systems should be developed only once we are confident that their effects will be positive, and their risks will be manageable.”

It’s time for Homo sapiens to pull the wagons into a circle and consult about ethical issues and formulate the policies which will prevent falling over the brink.

 Therefore, we call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4. This pause should be public and verifiable,  and include all key actors. If such a pause cannot be enacted quickly, governments should step in and institute a moratorium.

Regarding a general maxim to guide ethical deliberation broadly, I’m pondering the wisdom of a recent principle enunciated by Noreen Herzfeld. Dr. Herzfeld emphasizes that AI should be thought of as a tool to enhance human wellbeing. Just a tool. Only a tool. Not a person.

We approach the scary AI brink when we try to humanize AI. Why do we pursue AGI? Why do we manufacture sex bots? Why do we refine the chat box? Stop it! Herzfeld’s principle: let AI be AI, and let humans be human. Don’t interbreed them.

AI Brink #2: Malicious Malware

Is Silicon Valley providing bad actors with the computer equivalent of the AR-15s used by today’s mass murderers? American and Austrian gun manufacturers went over the brink. And now blood flows like rivers in our schoolyards.

Some armies have already launched autonomous lethal weapons. Once released, there’s no calling them back. And sometimes they kill untargeted people. Is this the future our AI techies wish for our planet?

Malicious actors the other side of the AI brink are ready and eager to pull the trigger to flood social media with fake news, re-edit distorting pictures, sway elections, and drain our bank accounts. Should the present generation of AI techies distribute the digitized equivalent of AR-15s to malefactors who are ready to rob, pillage, and destroy?

What we need are carefully constructed guard rails. The human geniuses of AI should unite with one another this side of the AI brink to construct those moral guardrails. That’s what Elon Musk, Geoffrey Hinton, and Steve Wozniak urge us to do before it’s too late.

AI and Faith at the AI Brink

Our Jewish, Muslim, and Christian colleagues at AI and Faith are accepting the Open Letter invitation to contribute to ethical deliberation and public policy formulation. The mission of AI and Faith “is to equip and encourage people of faith to bring time-tested, faith-based values and wisdom to the ethical AI conversation.”

In a newsletter article, David Brenner readies us for the challenge.

“Working together, we have a better chance to affect the larger AI ethics conversation by speaking in respectful, robustly pluralist voices, with no faith tradition toning down its faith beliefs, unified in our common contention that faith values of three quarters of the world’s peoples are essential to the debate. That approach has gained us admission as the only overtly faith-oriented partner among the 106 organizations in the global Partnership on AI.”

Brenner offers what I call public theology at work.

Ethical guard rails this side of the AI brink just might be what safety requires.

Conclusion

“AI has great power to do good and evil,” wrote Elon Musk in a Tweet. “Better the former.”

You betcha, Elon!

▓

Ted Peters (Ph.D., University of Chicago) is a public theologian directing traffic at the intersection of science, religion, and ethics. Peters is an emeritus professor at the Graduate Theological Union, where he co-edits the journal, Theology and Science, on behalf of the Center for Theology and the Natural Sciences, in Berkeley, California, USA. He recently co-edited Astrobiology: Science, Ethics, and Public Policy (Scrivener 2021) as well as Astrotheology: Science and Theology Meet Extraterrestrial Intelligence (Cascade 2018). He also co-edited Religious Transhumanism and Its Critics (Lexington 2022) and The CRISPR Revolution in Science, Ethics, and Religion (Praeger 2023). Peters is author of Playing God: Genetic Determinism and Human Freedom (Routledge, 2^nd ed, 2002) and The Stem Cell Debate (Fortress 2007). See his blogsite [https://www.patheos.com/blogs/publictheology/] and his website [TedsTimelyTake.com].

See his fiction thriller, Cyrus Twelve, with its transhumanist plot.

▓