A cursory glance at this year’s headlines suggests that “technology” might as well be synonymous with “artificial intelligence”. Even as many struggle to prove its value, business leaders can’t seem to get enough of generative AI, eager as they are to alchemise pilot projects into something revolutionary.

Spending on the technology is through the roof. AI has rocketed up the agenda for governments around the world, which increasingly view the tech as a matter of national security. The UK and the US, for instance, are developing ambitious sovereign-compute projects. Meanwhile, Silicon Valley firms are spending record amounts on AI infrastructure.

However, the technology has not come without controversy. Although more people than ever before are using AI tools, citizens are increasingly concerned about AI’s impact on jobs and society as a whole.

The race for AI skills

With AI supposedly coming for everyone’s jobs, from coders to poets, organisations are increasingly waking up to the need to upskill, well, everyone. For governments and enterprises of all sizes, preparing workers to use AI has become a priority in recent years.

For instance, Infosys, an IT-services firm, has introduced a layered upskilling programme, where employees receive training based on the level of AI expertise required for their job function. The programme classifies staff as AI “users”, “builders” or “masters”, and outlines different training modules for each group.

Meanwhile, in the Isle of Man, legislators are working to make AI training available to every citizen and business in the crown territory. And, across the UK, schools, colleges, community hubs and workplaces will soon offer AI-skills courses as part of a government-led initiative to train 7.5 million UK workers (about one-fifth of the workforce) to use GenAI effectively. Keir Starmer, the prime minister, has put AI at the heart of the government’s industrial strategy.

The UK’s AI action plan

When the Labour Party came to power in 2024, the new government recruited Matt Clifford, an investor and AI evangelist, to develop a blueprint for making the UK an AI superpower. Thus began the creation of the ‘AI action plan’, which was unveiled by Downing Street at the start of the year.

The plan will “mainline AI into the veins” of the UK, according to Starmer. Its flagship policies include loosening planning restrictions around so-called AI growth zones (clusters of data centres) and developing sovereign data networks and computing infrastructure.

Starmer also said that the UK must carve its own path with AI regulation, especially given that it now has the freedom to do so, post-Brexit.

Deepseek shakes world markets

When the Chinese AI startup Deepseek launched its R1 ‘reasoning’ model, a system that is free to use and cheap to train, the international markets took notice. Tech leaders had long assumed that LLMs require enormous amounts of energy and infrastructure to train and run, but the Deepseek model proved them wrong. Ripples were felt across the sector – the technology-heavy Nasdaq index plummeted by 3% – but worst affected was America’s largest chipmaker, Nvidia, which shed nearly $600bn (£484bn) in market capitalisation.

It wasn’t long before the company recovered, however, as market observers noted that if AI can be made cheaper and more energy efficient, demand for it would likely increase in the long term. Providers would simply seek to accomplish more with the technology rather than use less of it. Nvidia serves as a bellwether for the broader AI and tech industry.

The Deepseek saga failed to undermine the business models of any major semiconductor firm, although shares in Broadcom and Taiwan Semiconductor Manufacturing Co also initially slid on the launch. It may, however, have ignited an AI arms race between global superpowers.

The market shock brought to light China’s growing competency in developing AI models. Curious timing, given that Donald Trump, meanwhile, was preparing to launch a trade war intended at least in part to undermine competition against the US tech industry.

AI: the regulatory picture

The EU became the first major trade bloc to legislate against potentially harmful uses of AI. Its AI Act became law in 2024 and will be applied and enforced in stages. The legislation aims to regulate transparency and accountability for AI systems and sets out levels of acceptable risk for AI applications based on societal, ethical and legal considerations.

Like GDPR, the regulations apply to any organisation operating in the EU, so it’s not only businesses located in the bloc that are impacted. EU member states will enforce the act and non-compliance carries hefty fines. The worst infringers will face penalties of €35m (£30m) or up to 7% of annual worldwide turnover, whichever is higher.

This year, the first and second stages of the act have come into force. The first establishes unacceptable, unethical uses of AI, including for mass surveillance or discrimination. The second targets general-purpose AI platforms such as ChatGPT, establishing a broad range of compliance requirements for companies that use LLMs.

AI providers will be required to assess the safety of their models and demonstrate that they have done so through risk assessments and testing. They’ll also have to retain technical documents on model architectures, make those records available to regulators and publicly disclose training data if authorities deem it necessary.

Competing models

GenAI fans are always eager to try out new iterations of ChatGPT. Before launching its latest model, GPT-5, OpenAI stoked the hype flames by highlighting the LLM’s ability to give “PhD-level” responses, as well as its substantially improved programming abilities for would-be vibe-coders.

But some of ChatGPT’s most loyal users say GPT-5 lacks personality and offers generally unimpressive answers to prompts. OpenAI’s chief executive, Sam Altman, claimed that a malfunctioning model-switching feature meant that users were unknowlingly interacting with a “way dumber” version of the platform.

While AI advocates excitedly anticipate new models, some critics and sceptics have started to wonder how much further GenAI can be improved. If such tools have come as far as they can go, the long-discussed AI bubble could be set to burst.

Nevertheless, OpenAI’s popularity does not appear to be waning. The UK’s tech secretary, Peter Kyle, has apparently discussed a deal with Altman to provide a ChatGPT account to every UK citizen.

A cursory glance at this year’s headlines suggests that "technology" might as well be synonymous with "artificial intelligence". Even as many struggle to prove its value, business leaders can’t seem to get enough of generative AI, eager as they are to alchemise pilot projects into something revolutionary.

Spending on the technology is through the roof. AI has rocketed up the agenda for governments around the world, which increasingly view the tech as a matter of national security. The UK and the US, for instance, are developing ambitious sovereign-compute projects. Meanwhile, Silicon Valley firms are spending record amounts on AI infrastructure.

However, the technology has not come without controversy. Although more people than ever before are using AI tools, citizens are increasingly concerned about AI's impact on jobs and society as a whole.