Introduction to AI
Artificial intelligence is becoming very good and very fast. Whether music, text, code or images, the time when it was reliably possible to tell the difference between AI and human results is disappearing at an alarming rate.
The Double-Edged Sword of AI
But for all their magic, AIs can also be pretty useless. They make things up and misunderstand instructions. They are brilliant as toys, but incompetent as assistants. This makes it difficult to know how to express AI in the right light. Is it the most important technology trend since the iPhone? Or since the Industrial Revolution? It’s hard to tell from this distance.
Evaluating AI Intelligence
There are industry standards, so-called benchmarks, for evaluating the intelligence of AI models. These also show rapid improvement. When Google released its latest AI upgrade, it broke records across the board. But benchmarks are too narrow to provide completely reliable indications of skills and potential.
The Exponential Trend
To get a better understanding, it’s necessary to zoom out and look at the overall trend. When you do that, you see “a very strong exponential effect.” In an exponential trend, growth doubles and continues to double. At first glance, progress seems slow, but soon the line on the chart rises almost vertically. It’s a pattern you know from the COVID pandemic, which affected politicians and public health officials around the world.
Preparing for the Future
Now, AI is following a similar pattern, and there’s a worry that we don’t have a plan. The growth of AI could be as disruptive as COVID, depending on whether the growth continues and whether AI is good at as many things as it seems. If that were true, it would be much bigger than COVID. COVID was a temporary change, but AI will be a more permanent reshaping of the way everything works.
Reaching Peak AI
At some point, AI will stop getting better simply because tech companies are running out of energy to train their power-hungry models. The amount of energy it would take to train these models would be greater than is available on the entire planet. However, the data points out that AI capacity is doubling every seven months, and it is likely that this will continue for at least another five years.
Changes Ahead
There are still many changes ahead, some of which will be extremely positive, but inevitably, there will be disadvantages. As with any new technology, there will be a number of glitches. That’s why it’s so important to seriously think about whether this is true and what it would mean. We’ll get there, but we’ll only get there if we actually have a real plan.
Government Preparedness
In the short term, the government is actually doing a good job. They announced a government AI fund, which is a good thing. The UK’s AI Security Institute, which studies technical risks, is also a positive step. However, even the best AI forecasts are characterized by great uncertainty, and there could be a speculative bubble around AI. Nevertheless, the underlying technology is effective, and it feels like we’re on the cusp of something big.
