How AI is changing the way we watch F1

If you turn on the TV and flick over to Formula One, you’ll be hard pressed to miss the ‘insights’ graphics periodically appearing on the live feed.

Conveying sometimes complex information to viewers is a crucial aspect of motorsport broadcasting. After all, the sport is not always as accessible as others, especially with its data and tech-heavy terminology.

Netflix’s Drive to Survive opened the door to many new fans, but understanding the nuances of tyre degradation or pit stop deltas requires an entirely different level of engagement. This can be daunting at first, but the process of simplifying this information has been spearheaded by Amazon Web Services (AWS).

A global partner of Formula One since 2018, AWS has played a central role in the digital transformation of the series, and the results are already visible during broadcasts. Whether it’s timing and telemetry data, or something like the ‘Close to the Wall’ insight introduced last year, the cloud computing company is always looking to introduce new innovations.

Indeed, Formula One is a sport of fine margins. At the recent Austrian Grand Prix, eight-tenths of a second separated first-placed Carlos Sainz from last-placed Zhou Guanyu in the first part of qualifying.

There are over 300 sensors on each car generating over 1.1 million data points per second, and it falls to AWS to condense this data and present it in a way that is easy for the average Formula One fan to understand, all without diluting those dramatic, razor-thin margins on show.

Artificial intelligence (AI) is becoming an increasingly useful tool for simplifying that information and making the backend process more efficient.

Simplifying the process

One of the biggest challenges for AWS is ensuring that the insights it generates not only add context for those tuning into Formula One for the first time, but also entertaining for the diehard racing fan who watches every Grand Prix.

However, the ongoing evolution of AI technology is ensuring that time can be spent on evolving the offering, rather than on menial tasks.

“[Formula One’s streaming service] F1 TV […] includes some pretty interesting innovations like transcriptions,” Neil Ralph, principal sports industry specialist at AWS, tells BlackBook Motorsport. “This uses artificial intelligence to help a broadcast put up live commentary subtitles, historically a very human-based operation.”

Automating previously manual processes is a big focus for AWS, whose latest tool is ‘StatBot’, which uses generative AI to answer questions Formula One analysts might have about results sometimes dating back decades, such as the last time a rookie won a Grand Prix.

The old process saw race analysts take a question from the broadcast team and fact-check it against the historic data repository, find the answer, and then relay the information back to those on commentary.

That could take anywhere from a minute to ten minutes, which runs the risk of losing the editorial relevance of that specific statistic by the time it has been identified.

“StatBot is a forecasting assistant tool,” Ralph says. “It’s a pretty interesting way of using generative AI, which everyone’s trying to harness and take advantage of, and how to use that technology to help you mine large sets of data. Formula One started in 1950, now we’re in 2024.

“Using StatBot against that historic data repository, you get that answer back in seconds, so it’s available for that fan engagement aspect.”

AWS title sponsored the recent Canadian Grand Prix

Will AI be able to predict the future?

One of the current drawbacks of AI is that it can be very reactive in situations, rather than forecasting what’s to come. StatBot, for example, exists solely to provide answers to questions.

Another AI application being explored by Formula One and AWS is failure analysis. The collaboration sees generative AI applied to improve the efficiency of identifying and addressing technological issues that arise off-track during races. Known as ‘Root Cause Analysis’, it uses natural language processing to investigate system errors.

One such example could be when the timing ladder and telemetry failed during the Austrian Grand Prix, leaving the commentary team at a loss to explain the gaps between the cars.

“From an engineering point of view, [you’re] looking at trends and failures you’ve had in the past, or issues you’ve had,” says Pete Samara, director of innovation and digital technology at Formula One. “This will lead to prediction. We’re not there from a prediction point of view, but it will lead there.”

Go deeper

• How AWS is supporting F1’s sustainability drive using the cloud and AI
• The comeback trail: How Williams’ new commercial strategy is winning over sponsors

An example of what this programme looks out for is if a certain switch in a specific container has been failing. It’s not uncommon for a port to fail and for the relevant people to have no knowledge of it until the day comes that they have to use it.

So often in the past, the knowledge of certain failures was merely remembered by a particular member of staff. The faults would be logged, but trends would only be picked up by those that knew the quirks of the system. All it takes is for a specific engineer to leave the company and that knowledge is lost.

With AI, this problem is solved. But the next step is training the AI to proactively search for errors, rather than analysing trends and highlighting things that have failed in the past.

“I wouldn’t put a timescale on it [becoming proactive], but that is definitely the vision,” Samara explains. “We’re pushing really hard on phase one, which is [Root Cause Analysis], and we know that once we’ve done that, we actually understand how to use it, and understand all the things we can add into the proactive mindset.

“[Then the question is] how can a proactive mindset and proactive diagnosis of issues lead to preventative [problem solving]? We’re pretty aggressive, so it won’t be too long.”

‘AI will significantly improve the fan experience’

AI appears to be an increasingly important cog in the Formula One machine. But as Samara outlines, this is merely ‘phase one’. Ultimately, the question is how much further the technology can evolve – and how quickly.

Improving efficiency is the first step, but currently the technology is only used in a reactive way. For Samara, the goal is to make the technology “generic”. In other words, standardising its use across the business.

“We’ve got all stakeholders in the organisation energised to ask how we adopt this for the right reasons, but not to do loads of little things that don’t make a difference,” he explains. “We’re very active on prioritising and reprioritising [the right projects]. Something that was good a month ago may not be a good idea anymore.

“It’s always been about organisations looking for a vehicle to prove an idea, so the idea has to be firm. You have to stop it if you need to. So I think we’re pretty aggressive, and you’ve seen us working on two projects already and you’ll see more coming up.”

For Ralph, the focus is all on further developing a partnership that has already produced various innovations over its first six years.

“It’s a really interesting moment in time to look back at the six years we’ve had, because we’ve come from Formula One having a real absence of AWS capability, [but] it’s a proud moment to see they are embracing this in the right way,” Ralph considers.

“To try to look ahead to the next five years is super complex. No one saw generative AI coming 18 months ago, and now it’s revolutionised the way every industry looks at artificial intelligence.

“I think it’s very hard to extrapolate that far, but I have very little doubt it will significantly improve the fan experience in Formula One and the way that this partnership will operate in the future.”