Introducing the Human Collective Intelligence Alignment Problem: Say Hi to HI
/AI is not ‘Alone in the Universe’ (ELO)
As we witness the dawning of the age of artificial intelligence – a post-anthropocene or alternatively, the silicocene – it is easy to forget that AI is not the only form of intelligence in the world.
Long before the technology and even the idea of AI there has been HI: (carbon-based) human intelligence. More importantly, it is this ‘ambient intelligence’ and biological technology (i.e ‘wetware’, ‘human computers’) of HI or more precisely HCI – human collective intelligence – that is now creating – or in mytho-poetic terms, (un?)‘intelligently designing’ – AI in its own image and for its own ends. Given this relationship between HI and AI, a critical part of the process of aligning AI with moral values needs to be recursively ‘aligning its (HI) aligners’ (HI > AI).
‘Ghost in the Shell’ Mondo x SteelBook #018 (2017) from the japanese cyberpunk universe
An illustration of the immensity and seamlessness of the human-machine assemblage and collective superintelligence of civilisation as an emergent whole.
/‘The Winner Takes It All’? Or ‘The King that Loses his Crown’? (ABBA)
Currently, in our world of corporations and nations, artificial general intelligence (AGI) is being competitively developed, incentivised by windfall profits, (profit-maximisation) rapid economic growth (growth-maximisation) and great military power (power-maximisation).
Ultimately, a sufficiently advanced system could supercharge a corporation to monopolise the market or empower a state to secure world-domination. AGI has the potential to be an extremely powerful and transformative technology with extremely high stakes. For world-leaders and Big Tech CEOS, the silicon dream is that ‘the winner takes it all’ as the ABBA song goes (‘“Whoever leads in AI will rule the world” according to Putin).
An unfortunate side-effect of this ever-accelerating ‘rat race’ and ‘arms race’ is that the safety and alignment of AI systems are at grave risk of being sacrificed for speed and capabilities. According to the logic of the race to God-like AI, the faster systems can be made more powerful, the better: staying ahead of the competition and getting closer to ‘winning’ the game (of wealth and power) in the near-term matters more than safeguarding the long-term future and protecting the lives of all the people (human and non-human) that exist today. The problem with this mentality in relation to AGI is that nobody wins the race if it ends in existential catastrophe (perhaps, only the AGI). Rather than ‘the winner taking it all’, the neofeudal human elite of our time will be ‘kings that lose their crowns’ (to paraphrase another ABBA hit!).
/A 21st Century Tale of David & Goliath
As individuals, the powerful CIs that are developing AIs can appear as if they are already out-of-our-control. From our perspective, as (non-technical, non-governmental) citizens, AI development is effectively functioning like an automated process running without us, not needing our input (except our data) or asking for our consent.
Whilst fears surrounding AI tend to focus on the risk of AI Takeover – of an AI seeking power and disempowering humanity as a result– the real danger we face is in fact from the ongoing CI Takeover led by mislaligned tech corporations; that have lobbied against AI regulation or in the case of OpenAI, the creator of ChatGPT, value-drifted from their original non-profit structure to become a for-profit entity.
If we are already in the midst of ‘lock-in’, an unstoppable trajectory where one misaligned entity simply (intentionally or inadvertently) reproduces another, what chance do we have? Well, unlike a speculative rogue AI that functions as a ‘unified autonomous agent’ with its own monolithic value-system, the actually-existing CIs we see in the world today consist of many individuals with diverse, yet fundamentally human values. The researchers advancing AI capabilities are real people with real lives, with family and friends and hopes and dreams, all of which are imperilled by the AI race. In other words, there is a pre-existing misalignment between individual HIs and the larger CIs they are a part of that presents the possibility, in principle, for human values and agency to prevail over powerful economic, political and technological forces.
/ The Great Alignment: Aligning CI & AI as One
It is within our agency as aspiring aligned HIs to align the CI of humanity and AI as one. I call this time of prospects, as we escape the time of perils, The Great Alignment: a speculative future period of world history following ‘The Great Acceleration’ (the name economic historians give to the previous 75-year period of explosive development) where human, collective and artificial intelligence(s) and civilisational superintelligence as an emergent whole is aligned with the welfare of all sentient minds.
This vision of the future brings me great (existential) hope and sharing it brings me more still. We must remember that just as AI is not alone in the universe, neither are we. We are all part of a human collective superintelligence that, whilst far greater than our own, depends on the intelligence, agency and values of us all. In other words, we are not merely witnessing the age of AI, but actively constructing and steering it, together.
‘We are called to be architects of the future, not its victims’.
– Buckminster Fuller
