Three Big Ideas #19
Taking stock of tariffs, bridging the science-policy gap, and anticipating AI’s energy appetite
Welcome to our weekly Three Big Ideas roundup, in which we serve up a curated selection of ideas (and our takes on them) in entrepreneurship, innovation, science and technology, handpicked by the team.
⛔ Eamonn Ives, Research Director
Long before Donald Trump was returned to the White House, the world knew to brace itself for international trade tensions. On Saturday, the United States’ President made good on his promise to introduce fresh tariffs – levying an additional 25% tariff on Canada and Mexico, and a 10% one on China. President Trump has also declared that tariffs will “definitely” be imposed on the European Union too.
After rushed conversations with the leaders of Mexico and Canada, in which they agreed to do more to tighten border security and tackle illicit drugs production, America has since ‘paused’ tariffs against their countries for a month. While this will be a light reprieve, what’s undeniable is that we’ve been plunged into a new era of precarious mercantilism.
Amid the first skirmishes in the unfolding trade war, the United Kingdom seems to have emerged relatively unscathed. Though President Trump has accused Britain of being “out of line” on trade, he added that things “can be worked out.” Exporters here will no doubt breathe a sigh of relief at that news, and some might even sense an opportunity to supply attractively priced goods to the US market – at least relative to their European competitors if they likely end up burdened by tariffs. Should blanket tariffs simply cause the dollar to appreciate against sterling – as indeed it already is – that’ll, ironically, incentivise Americans to increase their appetite for imports.
With all that being said, the last thing I want to come across as is optimistic. Tariffs really are still a dumb idea. The Tax Foundation’s Erica York has estimated these new levies would decrease the size of the US economy by 0.4%. Beyond that, tariffs shrink global markets, and consequently make the entire world a less productive, less innovative, and ultimately poorer place. As we’re already seeing in this instance with China, tariffs are seldom implemented in isolation – countries tend to retaliate with countermeasures of their own, further pushing us into a trade doom loop. And while the UK may be getting a comparatively easy ride from President Trump for now, how confident can anyone be that he won’t simply change his mind tomorrow? I would forgive most British executives for being bearish in their reading of the situation.
Economists near unanimously agree that tariffs are bad. It’s good that Britain has so far dodged a bullet, but to think that makes us a winner would be most misguided.
🌁 Anastasia Bektimirova, Head of Science and Technology
Last year, I asked Dame Angela McLean, the Government Chief Scientific Adviser, about the biggest obstacles she saw in achieving a more productive relationship between academia and government.
Her reply was threefold. First, lengthy timelines. Research is unlikely to feed into policy in a timely manner if peer reviews take months, followed by more months for academic journals to publish accepted articles. Second, language barriers that leave the two camps lost in translation. Third, the culture of disagreement, which is a large part of academic training and ways of working. You can only get so far when deliberation focuses mainly on why ideas won’t work. Debate in academic and policy work is crucial, but you need to get things done at some point too.
Last week, UKRI opened a funding call to develop the policy-to-research infrastructure that would facilitate policymakers’ engagement with researchers and support evidence-based policymaking. The aim is to enhance public and civil servants’ scientific skills and their ability to access and apply evidence by increasing opportunities for engagement with the research community.
While the primary goals here are welcome, the valuable spillover effects that could occur are what most excite me. This could organically encourage systemic adjustments in academia, promoting more policy-relevant work and a clearer understanding of what research outputs the government actually finds useful – and how the academic system, including its incentive structure, needs to be adjusted to be more conducive to this.
For some more food for thought on the topic, here’s a perspective from Tom Kalil, CEO of Renaissance Philanthropy and former Deputy Director for Policy at the White House Office for Science and Technology under Presidents Bill Clinton and Barack Obama:
“If I were a university president, and I had a public policy school, I would want to give professors the option of having tenure and promotion based on real-world impact, not just how many highly cited publications they had. I wouldn’t mandate that, but I would make it opt-in. In many cases, when you do see faculty doing this type of work, they’re doing it in spite of, as opposed to because of, the incentive structures that they face. I think there’s a lot more we could be doing to encourage faculty to work on real-world problems.”
🔋 Jessie May Green, Researcher
DeepSeek, according to some, has given us AI’s ‘Sputnik moment’. Last week, word spread that the development of its R1 model cost a fraction of other notable equivalents, with fewer training hours creating greater efficiency. This sent shockwaves. DeepSeek swiftly unseated ChatGPT as the App Store’s most-downloaded free app, and panic ensued in the stock markets as investors questioned the valuations of other AI developers.
While bad news for competitor shareholders, some say DeepSeek may represent good news for the climate. Projections for AI’s energy needs are concerning. Some estimate that AI could absorb 75% of additional power in the US through to 2035, and President Trump is fast tracking fossil fuel power plants to meet the demand. In theory, DeepSeek’s efficiency brings hope that we can have powerful AI without ramping up greenhouse gas emissions.
In practice, this may turn out to be a fallacy. Though fewer training hours may reduce electricity consumption in the short-term, this would be unlikely to last. As explained by the Jevons Paradox, when a resource becomes more efficient to use, energy demands and costs do reduce, but then demand increases, and thus overall resource consumption goes up.
Thankfully, AI electricity demand projections may not be realised but ‘expect the best, prepare for the worst’, as they say. The projections only reinforce the requirement to decarbonise energy and ensure that power is clean at the source. If only nuclear fusion could get its boots on.
 | A guest post by
|