In the first of a new series of blogs, Adventures in AI, I discuss how there was nothing inevitable about the technological breakthroughs of the last eight years (bear with me. I am starting with the broader context before I focus on what this means for the chemicals industry). AI isn’t just another step along the road of computer science and the internet but is instead truly revolutionary. Like all revolutions, it so easily might not have happened. If you are going to fully understand AI’s potential, you need to start with understanding its history. Also see some thoughts below on what’s next for AI from IP battles to the potential for a US-China trade war.
By John Richardson
THE IDEA THAT artificial intelligence was an inevitable outcome of technological evolution is a common argument among sceptics. It suggests that AI’s current capabilities, from ChatGPT to DeepSeek, were simply a natural progression of the internet, computing power, and data availability.
But this view overlooks the critical role of contingent breakthroughs, financial investments, and strategic human choices in AI’s development.
A close examination of AI’s trajectory reveals that its rise was not a foregone conclusion. Rather, it was shaped by specific events, key research insights, and decisions made by individuals and institutions that could have easily unfolded differently.
The history of AI is not a straight line of inevitable progress, but a series of turning points, each dependent on funding, discoveries, and a willingness to take risks on unproven ideas.
One of the most defining moments in AI’s history came in 2017, when researchers at Google published the landmark paper “Attention Is All You Need”, introducing the Transformer architecture.
The importance of the Scaling Laws
This innovation changed the landscape of AI by allowing models to process language more efficiently and with greater contextual awareness. Without this breakthrough, the evolution of large language models like GPT-3, GPT-4, and DeepSeek might have been delayed by years.
Another crucial turning point was the realisation that scaling up models—both in terms of parameters and the amount of training data—led to dramatic improvements in AI performance.
The emergence of the Scaling Laws for neural networks demonstrated that model accuracy follows a predictable power-law relationship with increases in data and compute power.
This insight led companies like OpenAI, Google DeepMind, and Anthropic to invest billions into larger and more capable models. The success of these efforts was not inevitable—it depended on researchers recognising the pattern, securing massive computational resources, and convincing investors that scaling up was worth the enormous cost.
Financing played an equally critical role. Unlike the development of the internet, which benefited from widespread open innovation and government-backed research, AI breakthroughs required enormous private-sector investments.
Over the past five years, venture capital funding for AI startups has exceeded $290 billion, fuelling advances in machine learning, automation, and AI infrastructure. The US government has also increased funding, recently announcing a $100 million investment into AI research and sustainable technologies.
AI’s rapid progress is not the result of passive evolution but the outcome of aggressive financial backing and strategic decision-making.
US-China trade war risks and IP protection
Yet, even as the West continues to lead in AI, the emergence of DeepSeek—a Chinese AI model developed at a fraction of the cost of OpenAI’s models—suggests that AI development could be at another inflection point. DeepSeek’s R1 model was built for just $6 million, compared to estimates of over $100 million for GPT-4.
This demonstrates that the barrier to entry for developing competitive AI models is falling, potentially enabling more companies and nations to enter the race.
At the same time, DeepSeek also highlights one of the potential major weaknesses of China’s AI approach: government censorship.
Unlike its Western counterparts, DeepSeek is unable to answer questions about politically sensitive topics, reflecting the Chinese Communist Party’s strict control over information (El País). This raises concerns about whether Chinese AI models can compete globally if their responses are limited by government restrictions.
Beyond censorship, DeepSeek’s success has also raised questions about intellectual property rights. Microsoft and OpenAI are currently investigating whether DeepSeek improperly accessed OpenAI’s proprietary data or used restricted resources to train its model (Reuters).
If AI development becomes entangled in IP disputes, trade restrictions, and national security concerns, the competition between the U.S. and China may accelerate into a full-scale AI trade war.
No longer just about big models
The stakes could not be higher. AI is not just another technological innovation—it is a force multiplier for economic growth, national security, and demographic challenges.
With ageing populations in both the West and China, AI is seen as a solution to labour shortages, productivity stagnation, and healthcare efficiency. The country that leads in AI may reshape global economic power for decades to come.
In response to these challenges, US tech giants are not standing still. Meta’s AI chief, Yann LeCun, has doubled down on open-source AI development, arguing that open-source models will outcompete closed systems like OpenAI’s. OpenAI’s CEO, Sam Altman, has pledged to accelerate product rollouts and expand model capabilities to maintain leadership in the field.
And whatever the future of DeepSeek, its lower-cost approach to development is now out there, free to analyse. Has DeepSeek significantly lowered the barrier to entry to the degree that we will see a flood of highly innovative new entrants?
The AI race is no longer just about who has the biggest models—it’s about who can innovate the fastest while maintaining ethical and regulatory oversight.
AI’s progress was not inevitable. It was shaped by key discoveries, unexpected breakthroughs, massive financial investments, and strategic geopolitical decisions.
The story of AI is still being written, and its future will be determined by the choices we make today. The next inflection point may come sooner than we think—and it may not come from where we expect.