Too Good At Suboptimal Optimization
This morning's reflection: The fate of our world is disproportionately shaped by trade lawyers.
Trade lawyers are the key architects, coders, and negotiators behind most significant international agreements. Even agreements on areas that on first glance don't seem related to trade often stand to impact all sorts of different social and business interests. If the stakes are sizeable enough for a business or group of businesses, it then makes sense to engage a trade lawyer to help protect or advance those interests.
And many trade lawyers are remarkably good at what they do.
In fact, our problem may be that they are too good.
All things being equal, the objective of a privately-engaged trade lawyer will be to achieve the outcomes that best suit the interests of the parties that have engaged them. And if these are mostly private and business interests (which they tend to be, since good trade lawyers don't come cheap), then these interests tend to diverge from 'optimal' public interests.
Putting aside the debates of whether 'optimal' public interests exist, it's at least safe to assume that if private interests can afford to raise and invest substantial sums of money in trade lawyer fees, they are probably seeking agreements and outcomes that are certainly not whatever a 'public' would choose absent any private interests.
Too Good At Suboptimal Optimization
So here we are: an advanced professionalized civilization that finds itself too good at suboptimal optimization.
I'd like to outline two related musings in this post:
- PHAIs (Protocolized Human Artificial Intelligences), observing that the modern processes of professionalization and corporatization are effectively creating 'AI' behaviors through humans enacting the protocols of their corporate mandates and professional training.
- Public Actor DAOs, imagining DAOs that are created to serve as 'public interest' parties in this arena of agreements, organizing the resources to engage trade lawyers and align their optimization interests with public interests.
PHAIs
Modern culture is full of fears of AIs turning evil and going rogue. Paperclip-maximizing AIs turning the world into paperclip fodder, destroying our cities and forests in the process.
But wait. We're already doing all that.
Almost every major human civilization is guilty of large-scale ecocide.
Our corporations and governments comprise many smart (and perhaps even wise) people who yet somehow end up enacting collective decisions that are foolish on the planetary and ecological scales.
The AIs we fear are already here. They just run on humans, not semiconductors.
Corporations hire and mandate executives to maximize profits, ignoring externalities as far as profitably feasible.
Professions like lawyers train highly-skilled servants to live up to the ideals of client service (clients mostly being the institutions or parties that hire them).
Even with good people, this can't but lead to bad outcomes.
We can call these PHAIs, short for protocolized human artificial intelligences.
PHAIs effectively run and shape our world today. And they are also the foremost forces financing and directing the creation of today's fast-improving AIs.
If we can't redesign or redirect PHAIs, our AIs will do no better. If anything, they will just be too better at suboptimal optimization.
Never mind AI alignment problems, we must first face a PHAI alignment problem.
Public Actor DAOs
But how might we begin to address the PHAI alignment problem?
Let's go back to trade lawyers.
Trade lawyers are really good at what they do.
But they are trained (PHAI-ed?) to act in the service of those that engage them.
What if we could find ways to engage this skill and energy more in the public interest?
To an extent, this is the role of public prosecutors, attorney-generals, and government body commissioners.
But what if we could think bigger and more public? Could we imagine public commissions bigger than today's public institutions?
I think we can, and I think DAOs might offer us how.
Imagine a Public Actor DAO which can act as an engaging party to any major public agreement or negotiation (including being legitimately recognized at an inter-national agreement table).
This DAO could:
- Raise resources as a DAO does, through crowdstaking and other long-term bonds (imagine if you could buy public DAO bonds tied to the health of key ecological outcomes like river system biodiversity)
- Legally hold points and interests in a multi-stakeholder negotiation
- Define its interests for both the mid-to-long-term and case-by-case agreements through DAO mechanisms of deliberation and decision-making
- Engage and pay trade lawyers and other professionals to advance and protect these interests (perhaps with legal obligations and compensation clauses tied to these outcomes)
These might help fund and fuel a 'public' counterforce in all major negotiations.
It would also legally empower PHAI parties like trade lawyers and corporations to pursue public interest outcomes (instead of being legally bound to their professional client-serving duties, personal conscience or scruples notwithstanding).
Over time and scale, such an actor could begin to also partner and fund with the PHAI supply chain, helping raise and cultivate a new generation of public-actor corporations and professionals with richer and full-stack-public-responsibility mandates.
If something like this can begin to help realign PHAIs towards public interests (which I would extend to also include 'planetary' publics, not just humans), we might be able to harness our energies towards less suboptimal outcomes.