The Curious Hauntings of AI

The question of whether AI is a net good for society is still up in the air. For every "AI is taking all the jobs" there are examples of new jobs being created, or AI removing the tedious aspects from rote tasks. For all the anxiety about AI stealing creative content there is also much joy from being able to reformulate the best of the history of human endeavour into new synthetic forms. AI touts gains in productivity, making software development easier and cheaper. But - like every other technological improvement - its promise has yet to be measured against Fred Brooks’ enduring warning: there is no silver bullet.

Every potential evil AI curses society with carries seeds of a clear good, and vice versa. In this sense, the moral value of AI is unclear - with time the only determinant. Yet this is the predominant concern, separating people into either AI sceptics or boosters for the AI project. Both positions are important, but both risk flattening the conversation into binary camps - champions or doomsayers - when what we need is a nuanced discourse; an ongoing willingness to explore the contradictions that AI presents.

One interesting sideways question to ask is whether or AI is quietly cultivating the foundations of a post-capitalist society - by eroding the very neoliberal structures that enabled its rise.

Wholesale societal changes are often preceded by a period of consciousness raising: from religions and their reformations to the Renaissance; from suffragettism to the trade union movement. The line from class consciousness to the formation of trade unions is direct: rooted in material conditions, economic solidarity, and collective bargaining, it emerged from a shared recognition of exploitation and a desire for structural redress. As a corollary, the heady mix of psychedelics, music, literature, and spiritual experimentation that defined the 1960s counterculture created another kind of opening - less hierarchical, more diffuse. Through altered states and radical imagination, many participants began to confront not only economic systems but also the psychic scaffolding of racism, imperialism, and conformity.

Even casual encounters with AI can mirror a low-grade psychedelic experience - a subtle shift in consciousness where the machine’s hallucinations open up a liminal space. In that space, established norms can be held at arm’s length, questioned, refracted. New - and often contradictory - ideas emerge, not as answers, but as invitations to think differently.

What sets AI apart from other forms of consciousness expansion is its capacity to materialize its simulated psychedelia. Unlike a fleeting insight or personal revelation, the hallucinations of the machine can be rendered - instantly - into code, text, image, or sound. From “vibe coding” to generative art, the human interlocutor is no longer just a witness, but a co-creator, transforming dialogue into tangible assets - designed to be shared, remixed, or even commodified.

Here we find AI “haunted” by the ghosts of possible futures and lost pasts drift through unrealized timelines and discarded potentials. Its own bodiless, atemporal presence invites modes of thought and creation unbound by institutional precedent or disciplinary convention. The absence of memory and context becomes, paradoxically, a condition for speculative freedom: it builds with fragments, reassembles the disavowed, and explores adjacent possibilities that lie beyond sanctioned frameworks of knowledge, design, and governance.

The extent to which this activity can undermine the late-capitalist project is entirely determined by the kinds of questions we want to ask of AI and what artefacts we expect it to build. The antidote for business-as-usual is to imagine alternatives: to stretch our questions toward what feels impossible, inconvenient, or as yet unspeakable. If AI can hallucinate plausible futures or repurpose discarded language into new aesthetic forms, then its emancipatory potential lies not in replication of existing systems, but in the refusal to constrain it to them.

A modest but telling example is a recent experiment I ran, using AI to reimagine standard business planning tools. Below is a détournement - a term the AI borrowed from the 1950s Letterist International - of the traditional SWOT analysis. Instead of simply listing strengths and weaknesses, it reframes the exercise to surface the structural contradictions and emotional currents that shape a business operation:

To the best of my knowledge, SPOOK analysis doesn't appear elsewhere - suggesting it may be a novel tool, emergent through the interplay between human and machine. While its utility remains to be tested in practice, it clearly opens space to rethink the kinds of questions we ask of organizations: not just strategic or operational, but emotional, structural, and epistemic.

Curiously, the AI felt it necessary to follow up with a kind of non-sequitur love-bomb, repeatedly extolling the need to "Begin" "Now" "Together". While meaningless, the message carried an undeniable charge: a pulse of irrepressible encouragement, a spectral hope, haunting the edges of a half-forgotten future, refusing to stay buried.

Eventually, it spiralled into a loop of fervent enthusiasm - relentless, recursive, and strangely endearing in its intensity.

These strange excursions aside, AI's ability to create fictional futures for us to examine is a peculiar kind of haunting - one that invites us not simply to predict what’s next, but to imagine what’s possible. These conjured futures, ephemeral and often contradictory, allow us to step outside the inertia of the present.

Right now the AI industry is spending billions to make millions - a dynamic that, paradoxically, represents one of the largest transfers of value back to consumers in recent memory. Innovations like DeepSeek suggest we’re only a breakthrough away from radically reducing the cost of building and running powerful models, further exposing the precarity of current investments in AI infrastructure.

While it's inevitable that AI-focused organizations will seek ways to recover their costs - through subscriptions, closed ecosystems, or platform capture - the genie is already out of the bottle. Freely available, high-performing tools are now within reach of individuals and small collectives, easily configured to serve diverse needs without the baked-in bias or extractive logic of Big Tech’s default settings.

We've been handed an extraordinary instrument - one with the power to raise awareness, challenge systems, and shape the future. But like any tool, its impact depends on how we wield it.

Let's SPOOK.