AI helps to fix pot holes, but you need to get the basics right first

good morning. Back from tomorrow's holiday, Stephen created a neat acronym in internal politics last year: “PPPP.” For beginners, it represents a policy obstacle in the path to “patients, police, potholes, prosperity,” that is, the reelection of labor.

With the UK pitching to high-tech bosses at the world AI summit in Paris today, I wanted to see the pothole, one of the long-term pains that the government hopes AI can solve. Details of the following changes and challenges.

Opening up a new frontier

Potholes is not only fatal (cycling was killed or seriously injured in the UK due to road defects between 2017 and 2023). People have realised the true basics – and it's not just in the UK.

This 2017 survey of city elections in San Diego shows data across several election cycles showed voters punished incumbent leaders on tinsel roads.

The salience of that pothole in the real world is part of why Kiel's Starge defended what AI can do for “National Renewal” as his FT Op-Ed champion. Workers are turning their eyes to AI and working on the manifesto pledge to fix an additional million holes a year.

Some AI experts raised their eyebrows with inaccurate “feeding.” AI is not a sausage, it is a vast field of activity with a wide variety of applications. But that aside, can the council actually use AI to deal with potholes?

To answer that, it is worth understanding how we got here. On average, roads in all classes of England and Wales resurface once every 80 years, where the typical lifespan is about 20 years. If the road is not “recarbonized” frequently, it is likely to deteriorate and cause holes to squeal. However, due to the contradictions in the way Congress records data, the photographs of the government's national road conditions and how many potholes are filled are very patchy. Later, we will explain why variations across the country will explain this in issues of data governance and quality in AI.

In contrast to San Diego, where cities manage most of the local road maintenance in metropolitan areas, road maintenance in the UK is more fragmented, with different bodies managing different roads. The expressway and road A are located below the national highway. With the exception of the special cases in London, all other roads are maintained by local governments. Even the Greater Manchester combination authorities have each of the 10 district councils (Salford, Bury, etc.) administer their own highway networks. The council cannot raise additional funds to modify the roads if necessary.

Many councils hire private contractors to fix pot holes, but repairs are often not well-supervised to ensure high standards. A trench or repair that is inappropriately filled after the utility is working on can create a weakness.

Next, there are major factors. The boom in SUV sales over the past decade means greater pressure on the surface. European cars continue to get heavier due to the rise in electric vehicles, according to the European Federation of Transportation Environments (EVs are on average 23% heavier than traditional gasoline/diesel vehicles, due to the large battery). Met Office warns of wet autumn and winter more frequently, with more freezing and thawing cycles that tear the roads. This is a classic example of how early action is cheaper than waiting for things to come.

Drivers who compensate for pothole damage claims already cost the UK Council millions of people each year, and I was impressed that one highway division told the Asphalt Industry Alliance in its latest investigation into road health.

In other words, the council will need a predictable, long-term investment to switch from reactive pothole repairs to aggressive resurfacing.

This is where the government's AI plan could enter to identify potholes. Naturally, you might say, “You don't need AI for this. You can already see and report them to fixmystreet.com.” But as James O'Malley explains, humans easily miss the small tips that can graduate to a full-fledged hole. Councils should be able to identify potholes before anyone is dangerous enough to report it.

Rather than paying highway safety inspectors to go out twice a year, the council can take images of the streets with cameras and AI chips tailored to their vehicles, looking for holes and cracks, upload photos to the cloud platform, and passively construct maps of the area. It automates part of the inspection process and releases money to actually fix things. Comprehensive data means that councils can better prioritize maintenance by identifying potholes that are most expensive to fix if the council gets worse.

That's good, but it doesn't address the lack of resources and workers to address the flaws in the backlog and these new AI spots (after manual review).

Secondly, most technical solutions advertised by labour are Ten years old. In 2017, DFT worked with Thurrock, Wiltshire and York Councils to collaborate with trial technology by Gaist and Soenecs to photograph the vehicles “incorporated detailed images of the same section of the highway network” to “model degradation and ultimately prevent potholes.” These companies collected thousands of images (again, taken with the vehicle's camera) to train a model that assigns scores to the road section. The process probably used machine learning algorithms to classify images.

The exam was highlighted along with this 2019 Transportation Commission report and other advanced image processing technologies already in operation, but these things were not explicitly labelled as AI as six years ago, as AI was not so sexy. (At the time it was “analysed by a computer.”)

Workers' AI Opportunity Plans Tell Innovation as an answer. But that hard work took place years ago! The Surry Council, which currently uses AI only for testing, is proven. The problem we have is that digital transformation is slow and expanding across local governments where many authorities still store data in legacy systems that share real-time information between faculties (and different councils). Under these conditions, AI-driven pipelines (as good as the data they depend on) will never fly.

That's not just why detectors weren't successfully tried on councils a decade ago. The 2019 report covered the basic barriers that prevent the rollout. “If there's new technology… there's no central mutual way to share.” Also worrying: “The UK does not have the right incentives to maximize innovation.” why? These questions need to be answered before embarking on individual AI projects that require enormous investment and expertise to integrate.

It would cost more than 10 years and £16 billion to raise roads to reasonable standards in the UK and Wales. Until that becomes possible, AI has the potential to save efficiency. Risk arises when people expect more from “efficiency” than they can realistically provide. We have a lot of innovation and talented people doing cool things in this field. But the core issues remain. It's a lack of investment and a lack of long-term funding for large projects that will ultimately expand innovative technologies and ways of working.

Try this now

I celebrated the Vietnamese New Year. According to Hoa Viet's belief, this dessert is a common fixture for the annual east feast due to its sticky sweetness aimed at covering the “kitchen god” to ignore the flaws of the year.

You can hear me discuss one of my favorite childhood foods, instant ramen, in my new towker podcast series on Flavor Enhancer MSG and its cultural heritage (first time recording a pod in my kitchen!).

Today's top stories

• Second Labour MP apologises for WhatsApps | Keir Starmer was able to receive disciplinary action against a Second Labour MP over an offensive WhatsApp message following the looting of Health Minister Andrew Gwynne.

• Demand declines |Recruiters have reported the toughest conditions in the UK's job market since the Covid-19 pandemic, with no indication that they will regain employment confidence following Rachel Reeves' tax budget in October.

• The government claims it is being supported “Record the level” immigrant enforcement |Yvette Cooper will release a video this week of people being deported from the UK. The Interior Secretary attempts to blunt the rise of British reforms by claiming it is well-controlled against illegal labor.

• Peter Kyle's previous summit | AI tools are being used in the government to accelerate benefits claims, eradicate fraudulent garages offering MOT testing, and to stop complaints about false “Nimby” plans, I Paper (Paywalled) reports. Advances the global AI summit for political leaders and tech companies starting today, Science Secretary Peter Kyle has committed to “emphasizing people's hands” by making public services easy to access digitally. And he hopes the UK will lead the development of AI. “Tighten me – I'm still troubled by the lack of British deep seek,” he said.