The news around AI can seem relentlessly negative, and not entirely without reason. While some AI tools have proven genuinely valuable for individuals and businesses, there’s a genuine question around how much they help compared to how much they hurt. Creatives feel that their work is being used to train AI models that want to replace them, and a whole host of jobs in other industries.
However, the focus on the content generation part of AI makes it easy to overlook the potential benefits of other AI technologies. AI is increasingly being used for pattern recognition and data processing, with applications from medicine to the search for alien life. If AI can speed up the identification of substances from an image or sample, then, could this mean a revolution in identifying asbestos could be on the way—and potentially an end to asbestos-related diseases?
The different types of AI
When most people think of AI, they think of tools like ChatGPT or Gemini. Yet these content creation tools are far from the only areas where AI is being used. The same principles that allow those tools to learn from millions of images and pieces of text to piece together new content also allows them to learn from different kinds of data, and use that in different ways.
This is something that you might be aware of from driverless cars. Images from cameras around the car are processed in real time by AI algorithms, which then compare them to millions or even billions of similar images they’ve been trained on in the past. If they see something that they recognise as a risk, like an object moving in front of the car, they can automatically take the correct action they’ve been programmed for.
There could very well be an application of this technology to identifying asbestos. Asbestos identification currently relies on spectral imaging of samples to assess whether the material is present. However, this is a process that can take weeks, from the initial collection of the sample, to the analysis, and then communicating the findings to the client. You also have to be extremely cautious when collecting samples, both employing respiratory protective equipment (RPE) and thoroughly cleaning the site afterwards, given that there is no safe level of asbestos exposure.
The limits of AI for safety
It does seem plausible then that AI could help to identify asbestos. By training AI on images of asbestos at much lower resolutions than are currently required, you could speed this process up, and make it cheaper by not requiring such expensive equipment to analyse it. Asbestos can also be found in small quantities within an asbestos-containing material (ACM), something that AI image analysis might be able to identify more quickly than the human eye.
You might imagine that it could also assess the presence of asbestos without having to damage the potential ACM to take a sample, and risk exposure. However, this might be a stretch too far. It is possible that ACMs could be identified if enough images of them were fed into a model. However, it can be extremely difficult to conclusively spot an ACM by sight alone, even for a highly experienced assessor. There might be a suspicion based on the age of the building, and the apparent age and type of the material, but it can’t be conclusively proven in a way that guarantees safety.
Asbestos isn’t usually visible in a material, and the stakes of getting it wrong are so high that only conclusive evidence will really cut it. It’s also not something that should be encouraged, even if it proves useful. Take for instance the idea of sending in photos of flytipped waste to see if it contains asbestos. This might be a big help for councils to prioritise dealing with asbestos waste, but it would put members of the public at increased risk. Instead, it would be better to simply focus on preventing flytipping—and addressing all flytipping equally quickly.
Improving asbestos knowledge
Could AI have another impact though, one closer to the kind of AI most people are familiar with? There is still an enormous information gap when it comes to asbestos awareness, as many of our articles demonstrate. Many people are not aware that their houses may contain asbestos, as there is no legal obligation to mention it when buying a home. As many workplaces also do not tell employees when asbestos is present, this has led to a growing underappreciation of the risks of asbestos, which can seem like an issue of the past.
The reality is very different. The United Kingdom used more asbestos at the peak of the substance’s popularity than any other country, and thousands of tonnes of it still remains in place in buildings and objects across the country, as well as buried in brownfield sites. All of this is at risk of being disturbed and causing a new wave of asbestos-related diseases—and more so now that a wave of construction and refurbishment is happening in order to meet our net zero obligations.
It’s also a sad reality that too many businesses do not take asbestos safety seriously. This is partly true of recording and maintaining asbestos, where there can also be a lack of knowledge and little threat of inspections. But it’s also true of businesses that dispose of asbestos. Unlicensed asbestos removal remains an issue, while some contractors do not do their due diligence in checking for asbestos—and when they do find it, do not always remove and dispose of it safely.
Educating and informing
It’s an unfortunate fact that asbestos identification and removal can be costly (certainly more costly than ignoring it!). This is something AI might be able to help with, but that help only goes so far. As long as inspections are infrequent and information is lacking, this will continue. Where AI could help is by keeping people better informed. The more employees are made aware of the deadly risks of asbestos exposure, the more likely they may be to reject work, keep themselves safe, and force employers to act appropriately.
We’ve tried to do our bit towards this. Our UKATA Asbestos Awareness and IATP Asbestos Awareness courses are as low cost as possible, and easy to complete at home in your own time. But as long as there is any cost barrier, there will be limits to how many people take them. The availability of information for free online means that it’s vital that information is as accurate as possible, even if it isn’t as comprehensive as it needs to be to keep everyone safe.
This is where AI and asbestos could be a perfect match. AI seems to be increasing people’s willingness to ask direct questions, as it can provide more detailed and comprehensive answers than search engines could manage in a quick, digestible format. Given that, it’s important that the risks of asbestos aren’t minimised by AI text generators and search engines, and that we as an industry work to provide accurate information for these AI services to aggregate. That could make just as big a difference, if not more, to a future where asbestos is no longer an issue.