In the distant past before AI or even Google existed, when faced with doubt about how to behave, we used to ask friends, family and colleagues for guidance. For more serious or permanent matters, we might occasionally consult a doctor, lawyer or professional therapist.

We have now evolved into a civilization whose citizens typically spend half or more of their waking hours in front of a screen. Our lives are saturated with advice from multiple sources of authority. Politicians explain what policies we should support. Celebrities tell us what we should buy, believe and become. Scientists hired by tobacco companies spent decades telling us what we shouldn’t worry about. The word “should” is everywhere — and it almost never means what it pretends to mean.

When a politician says the country should move in a certain direction and that you, the engaged voter, should back their bill, they mean: This serves my interests, and I have painted it in your colors. When a brand-sponsored expert says you should feel reassured about some product or policy, they mean exactly what one of them dared to say out loud back in 1979: “Doubt is our product because doubt defeats the facts already in your mind.” Facts annoyingly prevent us from understanding the wisdom of the experts, who alone know what we should do.

Fortunately, some sources of true facts exist, or so we’ve been thought to believe. When editorialists of serious newspapers — even “newspapers of record” — explain how the government or the economy should work, they are advertising their own preferred ideology. If you share that ideology, you’ll most likely come away with a clearer idea of what other people should do, and thus dispense yourself from any undue effort. If you don’t share it, you should consult a different source.

The moral vocabulary of public life is largely a performance. “Should” is its favorite word precisely because it borrows the authority of ethics while remaining perfectly hollow. The word “should” is the workhorse of this vast industry. It is useful precisely because it borrows the grammar of ethics while committing to nothing. Whenever we hear the word should, we need to awaken our critical faculties.

A telling example with all the best intentions

In a recent article published here on Fair Observer, authors Farris Hamzeh and Natalia Hidalgo described the very real quandary that confronts an increasingly helpless Europe:

“Europe, meanwhile, has faced mounting domestic pressures and tests to its relationship with the US… Europe struggled to formulate a unified response to meet the moment. This initial hesitation gave way to a disjointed set of positions, with some European governments aligning with the US while others questioned the legality of US–Israeli strikes.”

These facts are painful for those of us who live and work in Europe and feel deeply concerned about its future. Yet, for all of us — including the authors of the article — it is often unclear what “Europe” even means, especially when talking about recommended public policy. Does it refer to the European Union? To Europeans themselves, many of whom have only a vague understanding of what the EU is, how it works and whom it represents?

Or perhaps “Europe” refers collectively to all the countries on the European continent, or at least to a supposed consensus among a majority of them. In some people’s minds, it could refer only to those seen as its leaders: especially Germany, France and the United Kingdom (which is not even in the EU). And what about the perception put forward by US geopolitical analysts who now frame the continent’s imagined divide as a struggle between “Old Europe” or “New Europe?” Talking about what Europe thinks and what it should do is by definition a perilous task.

This Devil’s Advocate always seeks clarity in the dossiers he studies. I need to determine who’s to blame for the confusion. Very objectively, I find this state of affairs concerning the very agency of Europe particularly inimical to formulating meaningful advice. The authors of the article, however, appear undaunted. In their concluding paragraph, they make a bold recommendation about what the phantom called “Europe” should do: “Europe should seek to drive a wedge between Iran and Russia, isolating the latter. Expediting the peaceful resolution to the standoff in the Strait of Hormuz is the best action Europe can take today.” 

It’s an intriguing idea, with unimaginably complex implications. To “drive a wedge,” you need a driver. Who is that likely to be? Their advice will likely appeal uncritically to readers whose worldview already treats the isolation of Russia as self-evidently desirable, a reflex deeply rooted in US policy since the Cold War. But, taking some critical distance — always a wise reflex — two important questions come to mind. Is it even feasible? And have we imagined the unintended consequences? As a tentative answer to the latter, I would submit that recent history teaches us that attempts to isolate Russia have a strong tendency not only to backfire but also to cause endless headaches for the perpetrators.

Had the authors used the word “might” instead of should, they could have gone on to address the difficulties of formulating and executing the policy they recommend. They could equally have begun evoking the longer-term consequences of such a policy. “Should” allows them to conclude their article with a sense of resolution. But does the reader feel that anything is resolved? And, quite frankly, will any of us Europeans act to take their advice?

Language, logic and moral force

This example should serve to help us reflect on how easy it is to create confusion with a word like should. When the authors say Europe should isolate Russia, they are recommending a concrete action. But when we say to a friend about to travel, “you should have pleasant weather in Athens in October” or “there should be an available room at the hotel,” we are literally predicting (probabilizing) on the basis of past knowledge. Such statements have no overt or hidden moral force. The auxiliary should seems to float between one extreme — Kant’s categorial imperative: “You should never lie, ever” — and uncertain but reasonably calculated speculation about what might happen in the future.

It’s this ambiguity that may lead us astray. And it’s against this backdrop that we must consider one of the more remarkable habits of our digital moment: millions of people, apparently unsatisfied with the quality of manipulation on offer from their politicians, celebrities, corporate scientists and editorialists, have decided in moments of doubt to ask an AI chatbot instead: What should I do?

The problem on the AI side is twofold:

• Chatbots are designed always to provide a “best” response even when there is no obvious one available, which is already an invitation to hallucination.
• The famous problem of sycophancy, or the tendency to approve everything the human prompter says, encouraging that person to persist even when danger signs indicating possible pathological behavior are present.

Numerous experts in AI behavioral practices have now emerged to warn us of the risks. Among the “5 practices to Avoid with Artificial Intelligence,” Professor Jairo G. Sarmiento Sotelo lists using “AI as a therapist or friend.” Applications that propose “therapy” with AI have recently emerged. Even the most sophisticated therapeutic chatbot “cannot understand the deep context of trauma, and it has no ethical or legal responsibility.” Obviously, an all-purpose chatbot is likely to prove even riskier.

What I say you should and shouldn’t do!

If you really wish to qualify for AI sanctity, this Devil’s Advocate makes the following recommendations:

1. You should use dialogue with an AI chatbot to explore two things:
   • original insights of “great ideas” you’ve never heard other people formulate that you think will help you better understand the complexities of the universe and human societies; 
   • nagging doubts you have about ideas and beliefs other people have persuaded you to adopt.
2. You should expect the chatbot 1) to flatter you, 2) to begin by offering the most banal, largely accepted explanations of the phenomena you’re interested in. Both of these signal a negative, uninspired beginning that it will be your job to move beyond and correct.
3. Don’t be fooled by its initial predictable response. Challenge it, even to the point of saying sarcastically, “that’s exactly what I expected from an AI bot,” and then find ways of breaking down its banality.
4. Never ask it, “What should I do?” Instead, ask it about who or what sources may provide you with further insight. It will actually help you find them.

I’ve called this the “sparring partner” approach, which I claim can help our society achieve true democracy to replace the simulacrum of democracy our overlords have bequeathed to us. It can also serve to improve our own mental health… so long as we keep sparring.

Finally, take this on board. I’ve just listed several things you should and shouldn’t do. I know nothing about your needs and ambitions. Therefore, you will be perfectly justified in taking none of them seriously.

*[The Devil’s Advocate pursues the tradition Fair Observer began in 2017 with the launch of our “Devil’s Dictionary.” It does so with a slight change of focus, moving from language itself — political and journalistic rhetoric — to the substantial issues in the news. Read more of the Fair Observer Devil’s Dictionary. The news we consume deserves to be seen from an outsider’s point of view. And who could be more outside official discourse than Old Nick himself?]

[Lee Thompson-Kolar edited this piece.]

The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.