Hello

This new report demonstrates beyond doubt what we already knew: that AI is the most important technology of our time and its social, economic, scientific and geopolitical impact is already profound.

If you don’t have time to read the full report, here’s our summary of the key points:

• 2023 was the “year of the LLM”. GPT-4’s release hugely surpassed all previous large language models, proving it could perform better than humans in applications such as the Uniform Bar Examination.

• This year has seen huge funding rounds, companies buying up graphics-processing units (GPUs) and battles around business models. NVIDIA’s market cap is now more than $1 trillion, as companies invest in AI infrastructure.

• Progress in AI capabilities has been rapid and profound. We have seen huge progress in self-driving cars, but also encountered significant questions around deepfakes and intellectual property as music and art can be created much more easily.

We are also opening new frontiers in biology, learning the rules of protein structure at scale and generating completely novel proteins. Google DeepMind’s recent AlphaMissense model, for example, uncovered the 71 million “missense” proteins likely to be the root cause of certain diseases.

• The surge in investment and attention has also made this the year that AI caught governments’ attention. Countries worldwide have shown huge diversity in their response; some have banned services like ChatGPT, some have introduced new frameworks and others are simply relying on existing regulation.

• “Compute is the new oil.” The US and China are engaged in trade battles and have imposed export controls on key materials, while countries such as Saudi Arabia and the UAE are making significant investments in compute and AI.

• Safety hit the agenda in a big way, with existential risks and calls for a “pause” capturing the attention of press and policymakers. While there is debate over how seriously and proactively we should be addressing these risks, many countries have begun acting to ensure they’re ready to meet the challenge. The UK acted first, launching the Frontier AI Taskforce and appointing Ian Hogarth as chair; the US followed in September by announcing the establishment of the AI Security Centre.

• The race for AI-enabled defence has started, particularly in the US and Europe. Defence technology has attracted record funding, with US defence startups receiving $2.4 billion in capital last year. NATO has launched a €1 billion innovation fund, while the European Investment Fund is understood to have allocated €200 million to defence investment.

• The need for international collaboration is well recognised, but proposals are mostly confined to papers. However, there are signs of progress, including the UK’s AI Safety Summit from 1 to 2 November, and the EU and US’s announcement of a joint AI “code of conduct”.

• AI companies are becoming more opaque about their research and architecture. As both competition and safety concerns grow, companies that were previously more open (even by name) are now quieter. This isn’t universal however, with Meta’s LLaMa launching an open-source race.

• As ever, the report made predictions for AI in the next year, including:

🔹 “A generative AI media company is investigated for its misuse during the 2024 US election circuit.”

🔹 “The US’s FTC or UK’s CMA [government regulators] investigate the Microsoft/OpenAI deal on competition grounds.”

🔹 “We see limited progress on global AI governance beyond high-level voluntary commitments.”

🔹 “An AI-generated song breaks into the Billboard Hot 100 Top 10 or the Spotify Top Hits 2024.”

TBI’s Take: What’s Next?

We’re still in the early days of AI, with breakthroughs occurring at pace. But the magnitude of what is happening already demonstrates why countries need to think about how they build competitive edge in this critical tech.

In our report A New National Purpose: AI Promises a World-Leading Future of Britain, we explored proactive steps that the UK can take to develop this edge, several of which have already been adopted. Many of these lessons will be applicable to, and essential for, countries globally, as the AI race accelerates and compute threatens a new digital divide for those left behind.

Governments need to understand, master and harness AI. They have always been slow to adapt to new tech, but AI presents a profound opportunity to reimagine the state and the services it provides.

Those who move fastest will shape the future.

ARTIFICIAL INTELLIGENCE

institute.global

View online