And yet—the selection of the term “hallucinate” this year in fact had little to do with climate, but rather due to the new meaning it has taken on with the advent of generative AI, a technology that has exploded over the last 12 months since the launch of ChatGPT in November 2022. It turns out that these AI chatbots also hallucinate—just in a different way from humans. Generative AI occasionally slips into fabricating information that has no correspondence to reality; a kind of misinformation that can be dangerous given the widespread and rapid adoption of the technology for everything from writing legal briefs to news articles. Without careful review and detection, these machine hallucinations are slowly working their way into our already frayed information ecosystem. 

Though the term has been criticized for overly anthropomorphizing the abilities of generative AI, I find it apt. Disinformation, in a way, has much in common with a hallucination; both phenomena distort our trust in our perception of the world around us. But while some “hallucinations” distort reality due to the misfiring of synapses or a poorly trained computer model, the distortions of disinformation, of the fossil fuel industry or otherwise, pollute by design. 

For years, climate action has labored under the shadow of doubt cast by the fossil fuel industry. This year’s COP seemed, if anything, worse than the previous summits; there were 2,456 fossil fuel lobbyists at COP28, outnumbering Indigenous representatives by 7 to 1. Even the CEO of Exxon (third largest oil company in the world by production), Darren Woods, made an appearance for the first time. The heavy presence of industry was not a coincidence; the hot-button issue at this year’s COP was the “phase out” versus “phase down” of fossil fuels. To an outside observer, the discrepancy in language may seem negligible, but in the world of climate policy, a few letters could mean the difference between meeting climate targets and blowing past them. In one sense, the heavy presence of the fossil fuel industry was a grim sign of industry capture; in another, it could be read as desperation. This time, big oil knew they had something to lose, like the rest of us.

The debate between “phase out” and “phase down” came down to a matter of scale and speed—how much would fossil fuels be curbed, and by when? The “phase out” language required a full reduction of fossil fuels by 2050; “phase down” prescribed no such timeframe. During the negotiations, the pendulum swung wildly between the two terms as the agreement was drafted and re-drafted. The final version landed somewhere in the middle. The text commits to: “Transitioning away from fossil fuels in energy systems, in a just, orderly and equitable manner, accelerating action in this critical decade, so as to achieve net zero by 2050 in keeping with the science.”

There’s no question any language about transitioning away from fossil fuels is a seminal achievement for COP. It’s also important to note that the bar for climate action at COP was in hell—it was, after all, the first COP agreement ever to explicitly mention fossil fuels. During the conference, U.N. Secretary-General António Guterres emphasized the importance of a commitment to a phaseout: “The 1.5 degrees Celsius limit is only possible if we ultimately stop burning all fossil fuels. Not reduce, not abate. Phase out.” The adopted language weaves around the commitment to a full phaseout entirely. What’s more, it includes a notable carve out for natural gas (using the coded term “transitional fuel”), stating that it can “play a role in facilitating the energy transition while ensuring energy security”—a long-time talking point of the gas lobby, despite the fact that, when accounting for leaky natural gas infrastructure, natural gas has the same warming potential as coal. Of the final text, Åsa Persson, Research Director at the Stockholm Environmental Institute says, “the outcome as a whole includes several conditions that may slow down the phaseout of fossil fuels.”

At the heart of the absurdity of climate politics—and, specifically, at COP28—is a strange kind  of doublethink that cleaves carbon emissions from the burning of fossil fuels. In an interview with the Financial Times, Woods complained that COP28 “put way too much emphasis on getting rid of fossil fuels, oil and gas.” Hearing Woods, one could imagine there is no link between burning fossil fuels and carbon emissions—the emissions just appear, with no particular source or cause (despite the fact that 90% of current CO2 emissions are due to the burning of fossil fuels). This rhetoric separating fossil fuels from their corresponding emissions is a disappearing act, absolving the industry of their responsibility for the crisis and protecting future profits. It also explains, in part, the tendency of climate policy to tackle demand, rather than supply.

This rhetorical separation also enables another step in the skewed logic of the fossil fuel industry: if emissions are the real problem, not fossil fuels, who’s to say that fossil fuels can’t be part of the solutions to climate change? In fact, who’s to say we need new sources of energy at all? To CNBC, Woods casts blame instead on the burgeoning renewable and green energy transition for the climate crisis: “I actually think that part of the thing that has slowed us down is this focus on making a step change and getting out of our existing energy system and starting something brand new. That is going to be a long, costly process that is going to be very, very expensive.” 

This language is pulled straight from a well-worn fossil fuel industry playbook—what scholars Naomi Oreskes and Geoffrey Supran call the “fossil fuel savior” narrative. This narrative promotes the worldview that climate change is “a downstream problem caused by consumer energy demand” and “insists […] upon society’s inevitable and indefinite reliance on fossil fuels.” In the words of Woods: “there’s demand for oil and gas today, there will be a demand for oil and gas going forward.” 

The savior narrative also positions the oil and gas industry as an “essential and inevitable part of the solution” to climate change, positioning renewables, by contrast, as unreliable and insufficient. Oreskes and Supran write that this framing allows the fossil fuel industry to “downplay its role in the climate crisis and to continue to undermine climate litigation, regulation, and activism.”

That is to say, in 2023, mainstream climate denial looks different. It’s no longer about bringing snowballs to the senate and asking why it’s so cold outside if there’s global warming going on. The new tactics are subtler. It’s asking: why all the fuss about fossil fuels when emissions are the real issue? It’s drawing fuzzy distinctions between “abated” and “unabated” coal. It’s carving out natural gas as a transition fuel. It’s caveating and loopholing the transition so that it comes too little, too slow, and too late. These careful changes in language produce a hallucination of a more subtle sort—not pure denial, but rather distortion of the future that we need for a habitable planet.

Only last month, the globe momentarily flitted over a dangerous threshold: average global temperatures were over 2 degrees Celsius above pre-industrial levels. It’s becoming increasingly apparent that while the 1.5 degrees Celsius target is still, theoretically, within reach, it is unlikely that we will meet it. Currently, the world is on course for nearly 3 degrees Celsius of heating. And yet, the prospect of losing the dream of 1.5 degrees Celsius is a near-impossible pill to swallow. 

In the original sense of the term, hallucinations have something in common with something else entirely: dreams. The connection between the two states of perception—and what they mean for our ability to discern reality from fiction—has been a subject of fascination to philosophers for millenia. Over 2000 years ago, in On Dreams, Aristotle drew parallels between the phenomenology of dreams and hallucination, arguing that they shared common roots. In the seminal Interpretation of Dreams, Freud expanded on the overlap between the two states of perception: “we may say, with every well-informed authority, that the dream hallucinates—that is, that it replaces thoughts by hallucinations.” On dreams, he writes, “we do not […] suppose ourselves to be thinking, but actually experiencing; that is, we accept the hallucination in perfectly good faith.”