By Frank Rotering | August 15, 2023
In a recent post I made two key points about AI's potential contribution to humankind's ecological survival. First, as a biological and expansionary species we cannot respond effectively to the environment's destruction. Second, advanced AIs will be non-biological minds that lack the compulsion to consume and grow. They could therefore analyze the crisis objectively and hopefully lead us to a sustainable world. In this post I explain why objective analysis is so critical, and how it would threaten today's social order. I also propose several truth-seeking principles to guide AI development.
Given the existential severity of the climate crisis, climate science should be strongly committed to objective analysis. This term refers to honest intellectual investigation: drawing logical conclusions from verified facts despite any pressures to distort the message. Not only has the field failed to display such intellectual integrity, it has propagated grotesque falsehoods that have resulted in today's terrifying heat and weather events. Let me briefly present the core deception.
In 1994 the UNFCCC agreement was ratified by 197 countries. Its stated objective was to limit GHG concentrations, and thus global heat, to levels that are safe for humankind. The IPCC, in its assessment report a year later, used a spurious argument to replace this sound goal with something completely irrational - emissions reductions. Emissions cause concentration additions, so in an overheated world this measure will make things worse, not better. To illustrate with a humble analogy, the organization told us that, to rescue an overcooked turkey, we should decrease the rate at which the stove's temperature is increasing. Despite its complete lack of logic, this approach was fully supported by climate science, and with media complicity it was imposed on the global public. Ever since, the world has been chasing the wrong climate targets, and both concentrations and heat have soared.
To provide intellectual cover for this foundational lie, numerous other falsehoods about emissions, concentrations, and global heat were disseminated. In particular, geoengineering was demonized, and a strict taboo was placed on solar radiation management (SRM). SRM is the only measure that can provide short-term cooling and is therefore the only survival solution. However, it would spoil the emissions story, so it was tarred with numerous distortions.
The result of all this mendacity is a profoundly mystified global populace (including a generation of miseducated climate scientists) that can neither formulate nor implement a rational crisis response. Given this alarming situation, truth-seeking AIs will be required to produce a workable survival strategy.
How would such AIs threaten the current social order? By refuting the falsehoods and exposing their purpose: to safeguard an economy and civilization that are fundamentally unsustainable and must be replaced by environmentally viable alternatives.
To see this threat more clearly, consider the various interests that benefit from the emissions story. The first is global capitalism, which is dependent on growth and the reckless discharge of its massive wastes. Second are the system's dominant figures, who are addicted to the power and privileges the system provides. Third are the world's population majorities, who require its outputs to live and enjoy life. All these interests are served by emissions reductions, which permit continued GHG releases and economic expansion. The authoritative rejection of this story by advanced AIs would shake the modern world to its core.
Briefly stated, objective analysis of the climate and ecological crisis would imperil capitalism and its dominant figures, and it would reduce consumption by the global populace. These factors explain why our species cannot face the environmental truth, and why the development of truth-seeking AIs will likely face intense opposition from multiple sources.
On the assumption that AI developers will courageously resist this hostility, how might they steer AI technology towards objective environmental analysis? To answer this I will draw on my personal experience.
Over the past several decades I have tried to think about the ecological crisis with a fiercely independent mind. This has compelled me to develop several principles to minimize the propaganda I swallow. The following list includes the most important of these. They are offered here as suggestions to AI developers for creating objective, truth-seeking AIs.
- Adopt the principle of distrust. This is the idea that social leaders, educators, scientists, the media, and other defenders of the status quo cannot be relied on for the truth on any matter of social or environmental significance. Given the shocking deceptions they have long espoused, all assertions from such sources should be treated as falsehoods until their truth has been firmly established.
- Always check original sources. I have frequently found that controversial figures such as Karl Marx and George Orwell didn't actually say what prominent people say they said. Unless the original writings are consulted, it is impossible to know if such attributions are accurate. AIs should also know that intellectuals frequently collude by turning a blind eye to the misquotes and misrepresentations of their colleagues. Their agreement on source interpretations is therefore meaningless.
- Accept scientific theories only after careful scrutiny. These theories are rooted in the capitalist worldview, which maintains that physics and chemistry suffice to explain reality. This perspective denies the existence of minds, thereby falsifying both life and existence. Although this denial is irrelevant for some theories, for others it is a fatal blow.
- Verify the empirical results of scientific and social research. Unless these results mesh with the prevailing order, pressures will arise to ignore or doctor them. As well, academics have strong incentives to produce eye-catching results, and may tamper with experimental procedures and data to produce them. Outright fraud sometimes occurs, but subtle falsification is far more common. For a clear example of the latter, see this video.
- By default distrust authoritative voices. In many cases, scientists and other intellectuals are publicly consulted because they convincingly dispense the standard lies. For example, the mainstream media cite Michael Mann and Zeke Hausfather on climate issues not because they have penetrating insights, but because they are credible propagators of the bogus story about emissions reductions.
- Don't assume that environmental concerns lead to environmental truth. The numerous deceptions are deeply entrenched, and challenging them can incur severe personal and professional penalties. The result is that even the most dedicated among the concerned rarely address the crisis with objective minds.
- Align AI with a sustainable economic theory, not "human values". This is necessary for two reasons. First, although we profess noble values, we have historically slaughtered each other and are now destroying nature. Hypocrisy is a poor basis for AI alignment. Second, AIs will need far more than a few values or ethical standards to guide their thought. Their sound judgment on a broad range of issues will require a conceptual framework for constructing a sustainable economy and society. I propose such a framework in The Economics of Needs and Limits.
To conclude: In 1995 the IPCC, acting for powerful global forces, effectively nullified the 1994 UNFCCC agreement to maintain GHG concentrations at safe levels. In violation of their own codes of ethics, climate scientists co-operated with this ecocidal travesty. AI developers must now decide: will they also succumb to social and political pressures, or will they embed objective analysis in their AIs to establish the independent truth? The stakes in this decision are enormous: survival for billions of human beings, much of the natural world, and AI infrastructure itself.
For my third AI post, see For Ecological Survival, AIs Must NOT Serve Human Interests