WC.com

Thursday, May 23, 2024

Arms Race

The British Broadcasting Corporation British Broadcasting Corporation (BBC) noted recently an evolving "arms race" in the realm of deep fakes. The article explains that some humans have a curious ability to identify fake photographs on a reasonably consistent basis. Despite that, the rest of us seemingly struggle in that regard.

The supposition is, therefore, that technology will be required. Algorithms will be trained to identify telltale indicia of artificial intelligence having created content. The result of this is likely to be a tit-for-tat escalation, wherein the software that creates, these photographs (movies, voices) will be persistently self-tweaking (learning) to produce better and better fakes.

Concurrently, the detection software will be self-tweaking and persistently evolving to become increasingly self-aware and effective at spotting them. As humans, we'll sit on the sideline, these various computer programs will apparently battle it out for supremacy as regards, our future, thoughts, and conclusions. We will be spectators in a vicariousness to rival some people's deep dependence on various sports, celebrities, and more. 

As I read, I was reminded of War Games (no, not the 2008 tweak/sequel/remake, 20th Century Fox, though that was not bad as sequels go). No, the real War Games (MGM 1983). It was edgier because none of us appreciated yet what a computer was or what it would be capable of. Sure, we had heard of computers in 1983. we were not New Yorkers for goodness sake. But, we had not learned enough yet to be scared of them.


In that MGM production, with the astute and aware David (Mathew Broderick) prompting and cajoling it, the evil computer (a WOPR, which they pronounced "whopper") taught itself the folly of "global thermonuclear war" by playing against itself in tic-tac-toe. Imagine a computer trying to outthink another computer, or worse to outthink itself. The premise was entertaining. 

We watched it play round after round to repeated, mindless, and inevitable draws. Despite knowing the probable outcome, we watched. In a movie setting, compelling and entertaining in the futility. In the NCAA, NFL, NBA, and more, perhaps not so much. People pay to watch the Browns, knowing full well how each season (perhaps largely each game) will end. 

But, is the reality of this new "arms race," any less preposterous. Can these computers compete with each other to any kind of supremacy or will they each simply tune, adjust, and persist indefinitely? I can just picture the screen flashing up that tic-tac-toe game at blinding and yet ever-increasing speed. Ultimately, might these competing manipulators and detectors simply stop and with my current conclusion
"the only winning move is not to play," War Games, 1983
As I strove to grasp that arms race prognostications conclusion, I ran across discussions on Quora and similar suggesting that, as computers, gain functionality, sentience, emotion, and greater capability, we must recognize their individuality, and afford them the rights and protections guaranteed under our constitution. You read that right, some feel your smartphone may one day have rights. 

In short, some contend that “all machines are created equal, they are endowed by their Creator (us) with certain inalienable rights.” And I pause. And I reflect. The courts have already started to side against the computers. AI-generated work is not capable of copyright protection. Can software, however sentient, therefore be guilty of defamation, or invasion of privacy? What would you do if it is convicted, put it in an isolation server somewhere to reflect on is miscreant errors, and rehabilitate? Would there be endless appeals to prevent powering it down?

Is this "the biggest bag of odorous excrement ever?" (Margin Call, Lionsgate, 2011) Are we at a moment in time when computers having rights is worthy of discussion? What rights would these computers have? Would it be appropriate for a computer to have the right to keep and bear arms? Does a computer have a right to privacy?

The astute reader will note that the United States Constitution in fact has no right to privacy. There are two points about that. First, the rights enumerated in the Constitution are certainly recognized. Second, that by no means equates to them being the only rights recognized by the courts.

Should computers and robots have the right to vote, hold office, be secure in their homes? Might we be eventually faced by a contest between two sycophantic automatons in an arms race of striving to entice voters with goodies while simultaneously striving to identify, expose, and leverage real or contrived flaws in the sentient program each is running against? It sounds terrifying, but perhaps so unrealistic it bears no further thought? Surely, we would see right through such contrivance, or would we need our own AI to help us decide how to vote?

My goodness, perhaps bolstering the voter roles for your favorite candidate could just come down to an arms race of who could deploy AI programs fastest? No need to attract voters, AI could just build its own. With the AI writing other AI, the explosion of eligible voters could expand exponentially with extreme rapidity

I return then to The Wizard of Oz (MGM 1939), and keep repeating in my head "Lions, and tigers, and bears, oh My!"

An arms race? Will there be computers to tell us what to buy? Will there be computers to warn us about the programs that tell us what to buy? Will there be servers that flood us with pandering information its algorithm thinks we will like, or to which we are likely to react? Will the sheep follow the tech? John Archibald Wheeler once noted
"We live on an island surrounded by a sea of ignorance. As our island of knowledge grows, so does the shore of our ignorance."
I am not sure which is growing faster, the island or the shore. And, I am unconvinced I want to know the answer. I asked my AI, and it replied simply: "that is a paradox." So much for helpful AI.