Summarize this content to 2000 words in 6 paragraphs in Arabic Good morning. Stephen — who returns from his holiday tomorrow — coined a neat acronym in Inside Politics last year: “PPPP”. For the uninitiated, it stands for “patients, policing, potholes and prosperity” — ie the policy obstacles in Labour’s path to re-election. With the UK pitching to tech bosses at the global AI summit in Paris today, I wanted to look at potholes, which loom large in the minds of many Britons and are one of the long-running sores that the government hopes AI can solve. More on the potential for change and the challenges below.Breaking new groundPotholes can not only be fatal (255 people cycling in Britain were killed or seriously injured owing to road defects between 2017 and 2023) but they stand out as things that a) the state is not good at fixing and b) are getting worse. People notice the real basics — and that’s not just in the UK.In this 2017 study of city elections in San Diego, data across several electoral cycles showed voters punished the incumbent leaders for shoddy roads: as the number of “pothole complaints” went up, the incumbent’s vote share went down. That salience of potholes in the real world is partly why Keir Starmer mentioned them in his FT op-ed championing what AI can do for “national renewal”. Labour is turning to AI to tackle its manifesto pledge to fix an additional million potholes a year:Some AI experts raised eyebrows at the imprecise wording of “feeding through” — AI is less a sausage, more a vast field of activity with diverse applications. But that aside, can councils actually take advantage of AI to deal with potholes? To answer that, it’s worth understanding how we got here, to a place where roads are being branded a “national embarrassment” by MPs. On average, all classes of roads in England and Wales are resurfaced once every 80 years, when the typical lifetime is about 20 years. If roads are not “re-carpeted” frequently enough, they become prone to degradation and potholes. But the government’s picture of national road conditions and how many potholes are being filled is very patchy, because of inconsistencies in the way councils record data. I’ll get on to why variations across the country in data governance and quality matter for AI later.In contrast to San Diego, where the city manages the majority of local road maintenance in the metropolitan area, road upkeep in England is more fragmented and different bodies manage different roads. Motorways and A-roads come under National Highways. Barring the special case of London, all other roads are maintained by local authorities. Even in the Greater Manchester Combined Authority, each of the 10 district councils (Salford, Bury etc) manages their own highway network. Councils aren’t entirely free to raise additional money to fix roads whenever needed.Many councils hire private contractors to fix potholes, but the repairs are often not sufficiently supervised to ensure high standards. Improperly filled trenches or repairs after utility works can create weak points.Then, there are big-picture factors: the past decade’s boom in SUV sales means greater pressure on the surface. Cars in Europe will keep getting heavier because of the rise in electric vehicles (EVs are 23 per cent heavier on average than traditional petrol/diesel cars because of their large batteries, according to the European Federation for Transport and Environment). The Met Office has warned of more frequent wet autumns and winters — leading to more freeze and thaw cycles tearing up the roads. It’s a classic example where early action is cheaper than waiting for the things to crop up. Compensating motorists for pothole damage claims is already costing English councils millions every year and I was struck by what one highway department told the Asphalt Industry Alliance in its latest survey of road health: “Our maintenance backlog is more than 10 times our annual budget, so we have to be very selective about what we pick and don’t pick to maintain.”In other words, councils need predictable, long-term investment to switch from reactive pothole repairs to proactive resurfacing. Now this is where the government’s AI plan for identifying potholes could come in. You might understandably say “we don’t need AI for this, we can already see and report them to FixMyStreet.com”. But, as James O’Malley explains, humans easily miss smaller chips that could graduate into full-blown holes. Councils need to be able to identify potholes before they become hazardous enough for someone to report it. Rather than paying highway safety inspectors to go out twice a year, say, to check the streets, councils can fit their own vehicles with cameras and an AI chip to take images, look for potholes or cracks and upload the photo to a cloud platform, passively building a map of the area. That automates part of the inspection process and frees up money to actually fix the things. Comprehensive data means councils can better prioritise maintenance, by identifying the potholes that will be the most expensive to fix if they deteriorate further. That’s good, but it doesn’t address the lack of the resources or workers to address the backlog and these new AI-spotted defects (after vetting them manually).Second, the tech solution being touted by Labour is almost a decade old. In 2017, the DfT worked with Thurrock, Wiltshire and York councils to trial technology by Gaist and SOENECS where vehicles take “regular, detailed images of the same sections of the highways network” to “model deterioration and ultimately prevent potholes”. These companies collected thousands of images (again, taken by cameras on vehicles) to train models that assign scores to road sections — and the process probably used machine learning algorithms to classify images. The trial was highlighted in this 2019 Transport Committee report along with other advanced image processing techniques already in play — except these things were not explicitly labelled AI because six years ago AI wasn’t as sexy. (Back then, it was “analysed by computers”.) Labour’s AI opportunities plan heralds innovation as the answer. But that hard work was done ages ago! Surrey council, which now solely uses AI for inspections, is living proof. The problem we have is scaling that up across local government, where digital transformation has been slow and many authorities still store data in disconnected legacy systems that make sharing real-time information between departments (and different councils) tricky. Under those conditions, an AI-driven pipeline — which is only as good as the data it relies on — will never fly. That’s not the only reason why detectors successfully trialled in councils a decade ago haven’t spread. The 2019 report explained the basic barriers that hamper rollout: “when there is new technology . . . there is no central reciprocal way that it can be shared”. Also worrying: “the UK does not have the right incentives for innovation to grow to its full potential”. Why? We need to be answering those questions before embarking on individual AI projects that will require huge investment and expertise to integrate. Bringing roads up to a reasonable standard in England and Wales will take a decade and more than £16bn. Until that becomes feasible, AI has the potential to make efficiency savings. The risk arises when people expect more from “efficiency” than it can realistically deliver. We have lots of innovation and talented people doing cool things in this area. But the core issue remains: underinvestment and a lack of long-term funding for big projects to enable us to finally scale up innovative technologies and ways of working.Now try thisI just celebrated the Vietnamese new year with helpings of bánh tổ, a steamed cake traditionally eaten for the occasion. According to Hoa Viet beliefs, this dessert is a common fixture of the annual feast because of its sticky sweetness that is intended to bribe the “kitchen god” into ignoring the household’s faults during the year.You can hear me discussing one of my favourite childhood foods — instant ramen — in a new Lecker podcast series all about the flavour enhancer MSG and its cultural legacy (my first time recording a pod in my kitchen!).Top stories todaySecond Labour MP apologises over WhatsApps | Keir Starmer could take disciplinary action against a second Labour MP over offensive WhatsApp messages, following his sacking of health minister Andrew Gwynne.Dropping demand | Recruiters are reporting the toughest conditions in the British jobs market since the Covid-19 pandemic, with no sign of employers regaining confidence to hire following Rachel Reeves’ tax-raising Budget in October. Government claims it is boosting immigration enforcement to “record levels” | Yvette Cooper will this week release videos of people being deported from Britain, as the home secretary attempts to blunt the rise of Reform UK by claiming she is successfully cracking down on illegal working.Peter Kyle’s pre-AI summit pledge | AI tools are being used in government to speed up benefit claims, root out rogue garages offering MOT tests and stop spurious “Nimby” planning complaints, reports the i paper (paywalled). Ahead of the global AI summit of political leaders and tech companies starting today, science secretary Peter Kyle has promised to “put power into people’s hands” by making it easier to access public services digitally. And he wants Britain to take a lead in developing AI. “Let me be blunt — I am annoyed that there isn’t already a British DeepSeek,” he said.
