OpenAI suggest a new Industrial Policy
One that will 'prepare democratic societies for the transition towards superintelligence'.
Open AI argue that market forces alone will not be sufficient to manage the speed, scale and risks of AI and that an ambitious, proactive public policy is needed.
They position the policy as being people first, democratic, adaptive and global (albeit US focused to start).
The guiding principles:
SHARE PROSPERITY BROADLY
AI should raise the living standards for all, not concetrate wealth and power. Through lower costs for essential goods, better health, eduction and economic security and partipaction in AI growth.
MITIGATE SERIOUS RISKS AT SCALE
Transition to superintelligence carries systematic risks including mass job disruption, cyber, biological and national security misuse, loss of human oversight or alignment, undermining of democratic institutions and excessive concentration of power.
DEMOCRATIZE ACCESS AND AGENCY
AI should become like literacy, electricity and internet access, something available to all. Affordable, agency to use and an ability to shape how it is used.
The recommendations are centreed around two pillars:
BUILDING AN OPEN ECONOMY
1. Worker voice in AI Deployment, at work, to improve job quality
2. AI first Entrepreneurship, make it easy for startups
3. A right to AI, a baseline economic requirement
4. Modernise the tax base, to reflect the income sources
5. Public Wealth Fund, to give every citizen a stake in AI growth
6. Energy and Grid Expansion, paying their own way with clean energy
7. Efficiency Dividends for Workers, pensions, healthcare, shorter weeks
8. Adaptive Safety Nets, Social Security, Healthcare
9. Portable Benefits, enabling workers to move jobs easily
10. Pathways into Human Centric Work, education, healthcare, community
11. Accelerate Scientific Discovery, scale discoveries faster.
BUILDING A RESILIENT SOCIETY
1. Safety Systems for High Risk Domains (cyber, biological, systemic)
2. An AI Trust Stack, auditability of AI actions
3. Auditing Regimes for Frontier AI, weight where controls are applied
4. Model Containment Playbooks, for when AI cannot be fully recalled
5. Mission-aligned Corporate Governance, balancing shareholder value
6. Guardrails for Government Use, how governing bodise deploy AI
7. Democratic Alignment Mechanisms, to shape AI goals and values
8. Incident and Near Miss Reporting, for learning and prevention
9. International Coordination, safe information sharing, crisis coordination
That is a lot.
Beneath this policy is fear.
1. That superintelligence will outpace the structures to survive it
2. The scale of job loss, work will have low meaning
3. Productivity gains benefit the few, consolidating wealth
4. Access becomes power
5. Corporations become more powerful than civilisations
6. Governments will misuse AI
7. Crisis without a playbook
SOURCE
Link: https://openai.com/index/industrial-policy-for-the-intelligence-age/
BESCI AI OPINION
This is not an 'Industrial Policy' it is a premptive defence strategy built by a PR team who are trying to shape the narrative.
It is clear that OpenAI assumes that wide scale job losses will happen, that wealth will be concentrated and that governments will be strained.
Rather than prevent these, which is too late as the open source code was made available early on, they are looking to 'shut the door after the horse has bolted' by
1. Redistributing some (but not all) surplus, retaining extreme wealth in those who benefit.
2. Building some safety nets which will likely benefit their 'safe' models, removing competition
3. Retain public consent, which will reduce legislative risk and keep their freedom to set the tone
The idea of taxing AI is ludicrous.
For generations corporations have optimised their structure to minimise tax payments and maximise benefits to their shareholders. The reality is that there will be less employment, and consumption taxes collected. Corporations are highly mobile and the technology industry is flexible, and not geographically constrained.
If they were serious then they would slow deployment, limit automation, spread or move (nationalise) AI ownership making a common asset.
It is PR
Is space the next frontier?
Bear with me on this one, it might be a far fetched thought.
With Elon Musk and Jeff Bezos investing in space, is this part of a bigger plan. Where AI data farms are in space, powered by solar, cooled by the atmosphere, using satelites to speak to earth.
What if you moved your tax basis away from earth to space, where there are no regulations, no laws, no governance.
Crazy thinking, right?
