Every generation believes it is living through a technological revolution. The steam engine transformed industry, electricity transformed civilization, and the computer transformed virtually every aspect of modern life. But something different is happening now. We are developing a technology that doesn't simply make human beings stronger, faster, or more productive. It increasingly has the capacity to think, analyze, create, predict, and make decisions once reserved exclusively for the human mind. Artificial intelligence is no longer a futuristic curiosity. It is becoming the infrastructure upon which the next civilization may be built. And while the world's most powerful corporations and governments race to control it, perhaps we should pause long enough to ask a rather unsettling question: Who, exactly, will control the future of humanity?
Let's break AI down into some smaller pieces and have a look. The first, and a very important piece to consider is the power problem. It has become common knowledge already that data centers are power hungry. They require copious amount of electricity to operate. If they were to be solely dependent on the existing power grid, there would inevitably be an increase in the light bills for ordinary consumers. Simple supply and demand. Even if the operators of data centers were required to contribute to the cost of upgrading and expanding infrastructure, the cost would still rise. The volume of data centers in the U.S. already under construction and in the pipeline are staggering. There are discussions regarding requiring data centers to supply their own power, with various options available to them. Natural gas power plants; dedicated natural gas turbines or engine powered generators, hydrogen fuel cells, small modular nuclear reactors (SMR's), solar power generation, geothermal energy, and large-scale battery systems. Combined heat and power is an option that deserves special attention. A conventional power plant wastes a substantial amount of heat. A combined heat and power system captures that heat and puts it to productive use.
For a data center, recovered heat could be used in absorption chillers to help produce cooling.
Think about the possibilities: generate electricity, recover waste heat, and use that heat to assist in cooling the very computers consuming the electricity.
It is an attractive engineering solution, although its economics depend on the cooling system design and operating conditions.
The concern is not simply that data centers will consume more electricity. It is who ultimately pays for the enormous expansion of generating capacity, transmission infrastructure, and distribution systems required to support them. If those costs are not properly allocated, ordinary consumers could find themselves subsidizing infrastructure built primarily to serve some of the world's wealthiest corporations.
The technology companies of tomorrow may not simply own the machines that process the world's information. They may own the power plants that keep those machines running, the infrastructure that connects them, and increasingly, the intelligence that operates within them.
Now let's move on to something more specific. Who owns the computing capacity, controls the algorithms, and determines access? We're talking about the potential concentration of economic, technological, informational, and even intellectual influence possibly being in the hands of a relatively small number of private organizations. Historically, the control of land, natural resources, manufacturing, and capital determined much of a society's economic power. In the emerging AI economy, control of computing infrastructure, data, algorithms, and access to intelligence could become equally consequential.
As far as control of infrastructure, companies such as Microsoft, Google, Amazon, Meta and Nvidia currently occupy important positions in this ecosystem. A new competitor might have brilliant ideas, but lack the capital, chips, computing capacity or infrastructure to compete.
Imagine a world in which businesses, universities, governments, hospitals, and individuals increasingly depend on a handful of AI platforms for:
- Research and scientific discovery
- Financial analysis
- Legal and professional advice
- Education
- Business planning
- Communications
- Medical information
- Engineering and design
- An investor in an AI developer
- A supplier of its computing infrastructure
- A distributor of its products
- A competitor developing its own AI systems
- Who determines which information is included in the search answers?
- Who decides which sources are considered reliable?
- Who establishes the rules governing responses?
- Who determines what information is restricted?
- Who audits the accuracy and potential biases of these systems?
- How transparent are those decisions?
- Computing capacity
- AI-powered business operations
- Data storage
- Cybersecurity
- Communications
- Financial and analytical services
- Military intelligence and surveillance
- Autonomous weapons systems
- Cybersecurity and cyber warfare
- Intelligence gathering and analysis
- Critical infrastructure and analysis
- Strategic economic planning
- Open source AI models allow developers to build on existing technology.
- Smaller companies are developing specialized AI systems
- Computing efficiency is improving
- Competition from China and other countries is increasing
- New semiconductor architectures may challenge existing suppliers
- Decentralized computing could eventually reduce dependence on large centralized facilities
No comments:
Post a Comment