AI Futures v1.0 · Part 1 of 2
Artificial Intelligence is increasingly woven into the fabric of society, bringing both transformative potential and deep systemic challenges. These advances also raise urgent questions: Who controls and benefits from AI, and at what cost to society and the environment?
Future Visions for Artificial Intelligence is a systems inquiry & speculative design project exploring a society with equal access and benefits of Artificial Intelligence.
The current AI ecosystem is system driven by exponential growth, concentrated control, and uneven distribution of its benefits and burdens. On one hand, artificial intelligence promises breakthroughs in productivity, science, and human capability; on the other, it deepens many of the systemic issues that already define our century: inequality, environmental strain, and the erosion of shared governance.
AI's rapid ascent is increasingly concentrated in the hands of a few tech giants that command the required compute, data, and capital.
Cutting-edge AI development today, especially "frontier" models – demands immense resources, both financial and natural.
By using their market position and ability to burn capital, they are creating barriers to entry for smaller players, which in turn creates further concentration.
A result of the massive deployement of capital is that skilled AI researchers and engineers from university labs are often hired by big tech companies, due to better funding and facilities. This leaves fewer experts to lead AI projects for good.
The promise of AGI was made with the premise of solving hard challenges. But to fuel the growth engine, the AI researchers end up working on creating easily monetisable products which require a lot of compute. For example, Sora 2.
If we ever achieve the promise of AGI, there will be a massive wealth and power transfer from the labour market to the capital market.
When reports talk about saving millions of dollars in costs because of AI; these savings are coming at the cost of salaries to workers. The AI services also are also provided by the few tech giants, so it essentially means lowering the wage rate, to benefit the shareholders of these tech giants.
shapingwork.mit.edu · Acemoglu, Macroeconomics of AI, May 2024
A part of the AI researchers believe increasing compute power exponentially is the way to achieve the promised goal of AGI. At the backend of this are massive data centres, which require land, electricity and water to run.
There are reports of prices of electricity increasing in the parts of the US because of data centre demand. The current scaling plans will require energy systems equal to powering mega cities.