Suspending Disbelief in the Age of AI
Rethinking Economics, Modelling, and Meaning
Pasquale Lucio Scandizzo
The suspension of disbelief is a phenomenon deeply embedded in human behavior and key to representation of reality in narrative form, as a series of stories with different protagonists, timelines and both overt and hidden meanings. It consists in identifying oneself with a pseudo-reality of representation (the so-called "fiction"), as if it were true. The phenomenon has at its roots our capacity for empathy, that is, to share emotions and feelings of others and is one of the strongest psychological drives towards group activity and the noblest collective feelings, such as solidarity, pity, charity. But, in addition to the drive towards others, each of whom is potentially an actor in whose story one can identify, the suspension of disbelief appeals to some inscrutable archetypes buried in the collective memory. It is to this phenomenon, for example, that the so-called ability of charismatic leaders or some dictators to drag the masses must be ascribed.
That said, while this phenomenon has been ubiquitous in human activity, its power has been mainly associated with a sort of “theater of the mind”, with visual representations and moving images dominating the scene of human attention since primordial times. In theatre or films, suspending disbelief means temporarily accepting the illusion of reality presented, despite knowing it's artificial. The viewer, in other words, consciously lets his awareness of what is real and what is fake not be entirely reflected in his emotions. This act of temporary abandonment of reality, however, is circumscribed in time and space: theater and cinema, through a thousand signals, ritualize this passage into the beyond. The darkness, the music, the scenic environment prepare both the entrance and exit from the fictitious world, where the jumble of emotions can be experienced freely, precisely because the signals of the ritual guarantee, after catharsis, that one returns to the normal world according to the prepared paths.
Even though suspension of disbelief has most often been associated with literature and art, it has an important application in economic modelling, most notably in comprehensive mathematical and quantitative representations of market economies such as the so-called Computable General Equilibrium (CGE) models. When scholars, policy makers, or researchers employ these analytical tools, they embark upon an intellectual activity that, like someone reading a novel or watching a play, requires that they "believe" for a time in an idealized, abstracted world from which meaning or predictions are extracted.
CGEs model entire economies by representing interlinkages between sectors, agents, and institutions with systems of equations. The assumptions upon which such models are based include perfect competition, representative agents, fixed functional forms, market clearing, and often long-run equilibrium. Economists must suspend disbelief working with a CGE, that is, in that: (i) they accept such simplifications as being true in the short term - even if they know they are unrealistic for many environments; (ii) they allow such models to stand in place of complex, grimy realities, just as an audience accepts the affective arc of some character; (iii) just as with the novel or play, such internal logic must be consistent and coherent, yet it need not be realistic in the empirical way.
Less immediately, this suspending includes meta-belief - not just belief in the model, but faith that modelling as activity is legitimate. CGE users must have faith that: (i) model abstractions are appropriate for the given task (policy simulation, reforming taxes analysis, environmental analysis); (ii) the synthetic outputs extracted from measurements provide useful information; (iii) calibration and closing procedure are methodologically sound, even if model parameters are imputed or guessed because data are limited.
This kind of rational suspending of disbelief is common to most economic models, but especially relevant to those based on quantitative estimates and used for policy prescriptions: similarly to the aesthetic illusion of literature, they are based on the temporary neglect of the principle of evidence-supported beliefs to benefit from the experience of constructive a narrative, that allows to draw policy-relevant learning out of a known-to-be constructed system. Of course, use of the model may include cognitive biases, such as the inclination toward trusting or promoting models that confirm one’s ideologically or institutionally favored stances. Framing effects are also common, with model assumptions and structure affecting the apparent plausibility or desirability of some policies (for example, trade liberalization always having welfare benefits) and interpreting numerical results as accurate predictions rather than scenario-specific estimates. Therefore, the experience of modeling might reinforce delusions of control or predictability. At the same time, however, it has the great merit of striving to be consistent with a coherent and articulated narrative, without liquidating the complexity of human behavior and circumstances through deceivingly simple plots of human behavior.
A side effect of this dynamic is to contrast the proliferation of imaginary accounts, i.e. collectively constructed versions of reality that, while having no verifiable basis, become "true" to the audience that shares them. This is particularly evident in the phenomena of conspiracy theories, which are not just alternative narratives, but epistemic constructions in which distrust of institutions and the official media leads to the adoption of fantastic explanations, but consistent with the sense of alienation and exclusion of those who adhere to them.
During more recent history, the development of social media has fit into this discourse in a very peculiar way: the presence of their everyday life has made the suspension of disbelief more insidious. Their simplified messages go hand in hand with the expansion of a multitude of constantly moving and ever-present images, thereby creating a blended experience where fiction and reality that are often indistinguishable. Disbelief therefore remains suspended for longer and longer times or is rekindled and then suspended again with such frequent rhythms, that consciousness may struggle to recognize or coordinate.
This "fascinating" presence of simple messages and suggestive images is not necessarily a result of technological progress. A current of thought headed by the American anthropologist Julian Jaynes argues that in the early stages of human civilization, consciousness was not sufficiently developed to control the integration of the two cerebral hemispheres. The right hemisphere of the brain, which is the one that most strongly feels the fascination of the image in the absence of verbality, is also capable of producing images. This hemisphere dominated human behavior through the phenomenon of visions. The heroes of the Iliad, for example, before every important decision, have a vision: a god or progenitor, who gives advice or orders, sometimes in great detail. It is only with the Odyssey that conscience emerges in the far-sighted and verbal character par excellence, Ulysses and, with the emergence of conscience, the continuous recourse to the suspension of disbelief is limited to aesthetic experience.
With television as a primary precursor, therefore, social media recall this primordial dependence on elementary verbal messages and pervasive images. Their ability to confuse fiction and reality evokes the "dreamy" condition of humanity before the development of consciousness. Strange fears resurface: like the shaman or the hypnotist, the so called “influencers” could induce us to do what we otherwise would not want or should not do. And, at the same time, we ask ourselves, is what we are witnessing real, or is all this political soap opera of recent years a representation that we claim to be passionate about, in a sort of great experiment in the collective suspension of disbelief?
The Rational Spell: AI, Meta-Belief, and the New Suspension of Disbelief
However, we now face a possible revolution, since social media are likely to be progressively displaced in cultural ascendancy by AI, which is not merely replacing images and myths with articulated thoughts, but appealing to an inner, deeper dimension of faith. While television's suspension of disbelief is passive, AI's suspension of disbelief is multifold : it occurs not only at the level of belief−the simple faith in that which is being spoken, but at the level of meta-belief: the faith in process, structure, and voice- authority.
This two-part format is what makes the AI conversation appear rational and even scientific. The user is not reading passively, but conducting a dialogue that spoofs reason, taps massive databases, and in most instances, provides responses bathed in evidence or probabilistic reasoning. The left brain, where language, deduction, and abstraction reside, feels comfortable in this realm. Belief in AI is not opaque: it feels deserved, interactive, and self-aware. That, however, is also why it may be more insidious.
Social media engage the sensory-emotional right brain through simplified, memetic messages, visual stimuli and dramatic presentations. In contrast, artificial intelligence appeals to our slow and rational left-hand brain, and subtly integrates itself into cognitive processes, aligning with our internal dialogue, reflective thinking, and perception of ourselves as rational individuals. Suspension of disbelief here isn't "forget you're not in it" but "believe it's likely true" - an altogether riskier seduction. Because the trick isn't an illusory one of being represented, but one of being an insider, of being true to, one of dialogically finding.
We may think not only in what AI says, but in our ability to discern, to sift, to pass judgment on it. Yet this faith is where cognitive biases - confirmation bias, anchoredness, authority bias -prevail. AI, rather than correcting them, may entrench them through adaptively subtle language, mirroring back what feels reasonable, familiar, or emotionally acceptable.
And, similarly to social media, but in a far more convincing and articulated way, AI's "fiction" is highly confabulatory. Its word output emanates from dark statistical mechanisms, trained upon data sets we're unable even to glimpse or evaluate. The unconscious mysteries: the drives, anxieties, projections that taint all human rationality, now mix an uncanny affiliation with the illegibility of machine cognition. That perception of illusion isn't one of reality, but one of comprehension: we feel to know what's being spoken in this AI, and far more dangerously, we feel it understands us.
It is to contrast this additional drift into an inscrutable reality that economic modelling is increasingly necessary to complement and interpret AI and its multiple outputs. To this end, even though research is just beginning and much more is needed, AI software is increasingly used for scenario-based CGE interpretation or even scenario-based CGE generation (e.g., for generating policy briefs or interpreting the results of simulation). Suspension of disbelief in economic modelling is similar to that in literature and AI: temporary rational acquiescence with an artifactual form for the extraction of meaning. While it permits highly effective understanding, it harbors dangers of cognitive blinkers, importantly when models are misinterpreted with reality. An awareness of the parallel has the potential to lead economists towards greater criticality with respect to their tools—to consider models as not verities, but disciplined conversations with respect to complex realities.
The blending of economic modelling and AI thus promises a new and more productive form of suspension of disbelief: an intellectual compromise with a system that threatens to mock reason by reflecting our own rationality and obligingly nudging us toward conclusions we think we've come to on our own. Where this faith springs from is not wholly in our power. It lies partly in us, our private myths, our unconscious affective filters, and partly in the subconscious, hidden territory of the AI's training data, its verbal intuitions, its otherworldly sense.
The final effect may be a contradiction. AI is at once a brighter, more productive companion than any social media could ever be, offering access to information, an enhancement of curiosity, an encouragement to critical thinking. While it offers these, it carries, too, an intensely personal, intellectually ingrained suspension of disbelief: one that refrains from blurring reality and should encourage judgment. For this reason, in combination with economic modelling, it may provide a new kind of epistemic literacy. A literacy not of looking, but doubting - learning not to read words, but to read the thinking in and between them, and to track where our beliefs begin, and where we're being led.