Far from being a far-flung Terminator-style future, artificial intelligence – or AI – is with us now and steeped throughout the automotive industry. We look at how AI is being used now in cars.
Artificial intelligence (AI) is something that not long ago felt like it was something that would happen in the distant future, but in reality, AI is being used across the automotive industry in many ways right now.
We’ve seen it start to creep into many aspects of our daily lives, such as Google searches or even on social media platforms, but now it is also widely used in many parts of the car buying and ownership experience.
While most of us associate the use of AI in cars as autonomous or ‘self-driving’ cars – limited to the likes of Tesla, luxury vehicles or not even yet ‘real life’ – it can be found right there from the beginning.
It’s there when the car is built in the factory, when you buy it in a dealership, when you get it serviced or repaired… it’s everywhere.
And that’s before you start to take into account the driving – autonomous or otherwise, AI in cars is being used for connected services too, for the vehicle’s safety systems and even in road traffic management and maintenance.
Let’s take a deep dive into how AI is used in our cars, and the benefits and challenges it brings.
How is AI used in cars?
Manufacturing and supply chains
AI-driven automation and robotics are being used in automotive manufacturing to make production processes more optimal and efficient.
While we all know robots have been making cars for decades alongside actual human beings, AI takes things a step further by looking at ways to reduce human error.
According to AI customer relations firm Salesforce, AI uses algorithms to analyse vast amounts of data to determine how assembly lines could be more efficient, as well as ensuring quality control – making them more productive and cutting costs.
One example of this is something like AI-enabled 3D printing, which, it says, allows for quicker and more accurate production of complex automotive parts, reducing the time and labour required for traditional manufacturing methods.
Mercedes-Benz, a leader in new automotive technology, uses AI in several ways, including using ChatGPT in intelligent vehicle production and to achieve energy savings in the painting process in the manufacturing of its next-generation MMA vehicle platform.
Another example can be found in how Mazda is tapping into artificial intelligence to cut the development time of its electric vehicles, as was reported by Drive in early 2024.
The Japanese car maker invested in AI-centric UK tech company Secondmind after entering into an agreement with the firm in late 2020.
Beyond this, manufacturers can also use AI for their supply chains, tasking it with things like analysing demand, monitoring inventory levels, and even watching social media, weather reports and market trends to assist with distribution, shipping and pricing.
“The automation and data analysis made possible by AI have significantly reduced the risk of human error and improved the overall efficiency and effectiveness of automotive supply chain management,” Salesforce says.
“As technology continues to evolve, we can expect even more advancements and improvements in this area, ultimately leading to a more seamless and streamlined automotive supply chain.”
Buying and owning through a dealership
Chatbots and virtual assistants have been around for several years now, though AI-driven assistants being used in dealerships is an area of growth.
Ask Harry is a technology offered by NSW-based software firm Contact Harald, and is essentially an AI-driven assistant designed to reduce the administrative workload in dealerships by streamlining service bookings and customer interactions.
The tech is integrated with what is known as Dealer Management Systems (DMS), using data to provide real-time analytics and detailed reports on customer behaviour and support dealerships by driving sales.
It’s the first of its kind in Australia and has in recent months been rolled out through a partnership with Suzuki on its ‘Hey Suzi’ hotline at all 92 dealerships nationwide.
According to Co-Founder and Director of Contact Harald Nick O’Halloran, Hey Suzi provides a convenient way for customers to book a service, make an enquiry or be transferred to roadside assistance if needed.
“It’s very hard at dealerships right now, they don’t answer the phone very often – and when they do answer, I think it takes eight minutes for the average person to make a service booking,” he told Drive.
“The advantage is that Hey Suzi answers the phone literally within two seconds, 24/7.
“The dealerships are so busy with the amount of phone calls that they’re getting, they can’t actually handle it. This way, we’re helping the dealerships and the customers by connecting them.
“Some people have asked if a service manager’s going to lose their job and so forth, but no, it will help them focus on bigger, more meaningful tasks.”
O’Halloran also said that, beyond service bookings, the technology can also organise a test drive for a new model and let people know when there’s a sale on that they might be interested in.
“So it will give you a call and say something like, ‘I noticed you made an inquiry three months’ ago – are you still looking buying a car?’. And you might say, ‘Yes, I’m still interested in the Suzuki Swift’. Then it will say, ‘How about we send you a brochure, and would you like us to organise a test drive for you?’.
“We can book that test drive right there with the customer rather than a sales person having to call you back and find out when you are available,” said O’Halloran. “It’s just making that booking system a much more efficient way of doing it.
“It could also give you a call to say, ‘We’ve got a Black Friday sale on [for example], 10 per cent off all cars, please come to your local dealership’.”
When asked by Drive if he felt this could be invasive for drivers, with the potential for advertising spam, O’Halloran said users can opt out of marketing phone calls if they wish.
“The options are there. Suzi is trying to make life easier for customers by helping them with whatever it may be, but if you don’t want that, then that’s fine. You can say you’re not interested and you want to opt out and that’s entered in the database.”
The capabilities of the technology, he tells us, don’t stop there either, with vehicle owners able to access help on things such as finance, roadside assistance, warranty information and even recalls – with people able to get an in-car notification to tell them there’s an issue with the car and organise a time to get it fixed.
At present, Suzuki is the only car maker in Australia using Ask Harry; however, according to O’Halloran, there is a lot of interest from dealers, especially those that have multiple brands under one roof and have seen how it works.
Hey Suzi will be a “game-changer” for the admin teams in dealerships, said Suzuki Australia’s General Manager, Michael Pachota.
Mercedes-Benz also told us it uses AI for admin tasks, as well as forming part of its IT infrastructure.
In the new CLA, to be revealed globally this March, its AI-powered MBUX Virtual Assistant will demonstrate how it is taking the “driver-car relationship to a new dimension with natural, human-like interaction, including empathetic characteristics that adapt to the driver’s style and mood”.
Connected services
Of course, it’s not only Suzuki playing in the AI space. Many manufacturers now have phone apps that offer connected services, personalising infotainment, providing real-time traffic updates and the like, seamlessly integrating your car and your smartphone as well as other devices.
These include Mercedes me, Kia Connect, Hyundai Bluelink, FordPass, myToyota, to name but a few.
Vehicle repair and insurance
AI is also already playing a role in how cars are repaired when they break down or are involved in a crash, with experts from the repair industry telling Drive it’s a complex area where the technology doesn’t necessarily improve anything.
Carly Ruggeri, Director at Euro Panels in Geelong, Victoria, said that she is encountering problems with AI being used by insurance firms for vehicle assessments to cut costs.
“For AI, efficiency is always the biggest argument, and insurance companies now use their own quoting methodologies and platforms, but it shouldn’t take over the actual thinking of a person because every repair is different,” she told us.
“Using a blunt instrument to say that this is exactly what you need every time is dangerous. That’s the aspect that we’re worried about, and it’s happening now to body repairers where it’s that ‘take it or leave it’ situation.
“It’s not about being anti-technology, but maybe having some policies in place where the accountability needs to be on the user.
“It’s all fine that the computer has told you what to do, but you actually have to use that human element to verify and be accountable for the end result – because the worst-case scenario is that someone gets injured or dies from a repair that hasn’t been done properly.
“It just can be really dangerous. It is oversimplifying and trying to replace the human element because of staff shortages; the answer isn’t just to cut people out.”
According to Ruggeri, many workshops are having to swallow costs they can’t afford because the AI being used by insurers is telling them to carry out jobs for less, and even when repairers do challenge this “nothing ever happens, it just goes into the abyss”.
RELATED: Mercedes “ready” to cop the blame if autonomous cars crash, as nations introduce liability laws
Matt Hobbs, CEO of the Motor Trades Association of Australia, said his organisation has just released a consultation on a new code of conduct between repairers and insurers because there are “already issues … where crash repairers and insurance companies don’t always agree”.
“AI does a first check and might say, for example, ‘Yep, that’s a three-quarter panel’. That makes sense. But it might also say, ‘You’ve got too many hours of labour, you’ve got too many parts here’. And that assessment process is where it often goes wrong,” he told Drive.
“That’s kind of the root of a lot of the disputes because repairers will say that job takes longer to do than what the AI is saying, or the method it is recommending isn’t the one recommended by the car company.
“What we’re trying to work out is a mechanism by which when they don’t agree, how do we deal with it swiftly?
“That’s something we’re grappling with at the moment because no longer is it the case that an assessor will come out and visit a shop.”
On the car maker side, in 2023 Porsche revealed it had partnered with US investment firm UP.Partners to launch a new business using AI to help mechanics diagnose and repair cars.
According to a Reuters report at the time, the start-up, named Sensigo, said its platform will eventually be able to predict problems before they occur reducing costs for customers while also improving profits for the business.
Safety and advanced driver-assistance systems (ADAS)
All new cars coming to market now must have some form of ADAS in them, because certain features are required by crash test authorities such as the Australian New Car Assessment Program (ANCAP) or its affiliated body Euro NCAP to get a five-star safety rating.
While ANCAP ratings aren’t legally required for a vehicle to go on sale in Australia, high safety scores have become a consumer expectation in recent years, and some safety features are now a legal requirement, such as autonomous emergency braking (AEB).
ADAS systems monitor the vehicle’s surroundings, detect potential hazards, and provide alerts or even take control in critical situations to increase safety and reduce road trauma.
The technologies assessed by the likes of ANCAP according to its current 2023 protocols include: car-to-car, car-to-pedestrian and car-to-cyclist AEB systems, lane support systems (LSS), automatic emergency steering (AES) and speed assistance systems (SAS).
From 2026, however, ANCAP plans to evaluate not just whether a car has assisted-driving technology but how it actually performs in its testing. The new protocols will assess factors such as driver engagement, vehicle assistance capabilities, and safety backup features.
“A range of modern vehicle safety systems leverage machine learning and advanced data analysis, which can be categorised as Artificial Intelligence (AI),” said ANCAP.
“For example, collision avoidance technology integrates data from multiple sensors to assess the environment and determine when to apply emergency braking or evasive manoeuvres. Similarly, fatigue detection systems monitor driver eye movements, blink rates and driving behaviour to identify signs of fatigue.
“These AI-driven technologies have significant potential to reduce road crashes and improve safety. However, their effectiveness relies on the quality of sensor inputs, the sophistication of data analysis and the thresholds set for intervention.
“As technology continues to evolve, ensuring the reliability and accuracy of these systems remains a key focus for ANCAP.”
Traffic management and road maintenance
As well as voice assistants, apps, and anti-collision technology in our cars all involving AI, there are also ways it is being used to improve traffic flow in our major cities, and even to repair broken roads.
According to Transurban, which operates toll roads in Melbourne, Sydney and Brisbane – as well as some in the US and Canada – its use of AI is not just something it plans for the future; it’s very much in the here and now.
For example, it’s being used in the tunnel in Brisbane which connects the Clem7 and Legacy Way tunnels by measuring and understanding when congestion builds up at exit points and determines whether to lower speeds further back.
On Melbourne’s CityLink, AI is automatically detecting when a vehicle has broken down and will flag it even before the humans – who are monitoring the cameras 24/7 – have seen it on their screens.
RELATED: Warning system at crash hotspot cutting accidents by 20 per cent
“We’re using AI to get our customers where they need to be faster, by helping us to detect the more than 1000 events that occur on our roads each week, including more than 360,000 stopped vehicles on Melbourne’s CityLink alone in the last year,” said Chris Jackson, GM Customer Experience and Operations.
“The benefits that our customers are seeing from this is demonstrated through a new feature in the Linkt app, which calculates customers’ estimated time savings using historic travel time data on their tolled trips and likely alternative routes.
“The data shows that nearly all drivers are seeing positive time savings each month, with median monthly savings reaching 52 minutes per customer, and one in five customers saving 150 minutes – that’s 2.5 fewer hours stuck in the car.
“We’ve also worked with Google to provide information for customers on toll prices, and our Trip Compare tool on the Linkt website helps them make an informed decision about using our roads.”
While the Transurban’s breakdown crews are out on the roads, their vehicles are also using AI too – scanning surfaces to determine where there are imperfections that need attending to to stop them from becoming potholes and the like.
In a sort of ‘two birds with one stone’ scenario, the cameras on the trucks are doing all this while the crews are otherwise occupied with breakdowns, so there’s no need for them to look for problems themselves.
What is the future of AI in cars?
Autonomous driving
We often hear a lot about autonomous driving, and, in particular, Tesla’s claims its vehicles are already capable of fully driving themselves.
However, the reality is quite different. While there are many things our cars can already do by themselves as it were, there is also a long way to go.
Automated vehicles can perform the entire driving task on a sustained basis without human input, either in all conditions or in specific conditions – and there are five levels ranging from mild driver assistance to fully autonomous.
| Level | Name | What it does | Examples |
| 0 | No automation | Warnings and momentary assistance | Lane departure warning |
| 1 | Driver assistance | Steering and brake/acceleration support | Lane-keep assist or adaptive cruise control |
| 2 | Partial driving automation | A combination of steering and brake/acceleration support | Lane-keep assist and adaptive cruise control at the same time |
| 3 | Conditional driving automation | Vehicle drives itself under limited conditions, user must still intervene as appropriate | Traffic jam chauffeur |
| 4 | High driving automation | Vehicle drives itself under limited conditions, no expectation for user to intervene | Robotaxi with limitations |
| 5 | Full driving automation | Vehicle can drive itself under all conditions with no intervention | Robotaxi with no limitations |
Refined for Clarity and International Audience.Current Australian laws do not allow for the use of automated vehicles on public roads.
The Federal Government, together with the National Transport Commission, are working on a new law called the Automated Vehicle Safety Law (AVSL), which will set out how such vehicles can be safely used on our roads.
A consultation on the new legislation was held in 2024, with the feedback now being analysed. However, limited trials of driverless vehicles have already been conducted on local roads.
In the future, this could enable the use of driverless ‘robotaxis’ – which do not have humans behind the wheel – similar to test programs being conducted by General Motors and Google in the US.
TV presenter Andy Lee shared his experience of getting into a Waymo robotaxi on a recent trip to Los Angeles, where a limited number of such vehicles are permitted to transport people around the city.
As it stands today, Tesla Autopilot and Enhanced Autopilot modes are legal in Australia and available to local Tesla owners, but Full Self-Driving (FSD) Beta is not.
The basic Tesla Autopilot comes standard with all new Teslas in Australia, while the Tesla Enhanced Autopilot is an optional extra.
While Tesla Full Self-Driving (FSD) Beta hardware is installed in all new models locally, you can’t use it yet. When this is allowed, it will be enabled via a software update.
Tesla has previously insisted that FSD does not make the vehicle completely autonomous and requires “a fully attentive driver who is ready to take immediate action at all times”.
“[Rolling out autonomous capabilities] should be a step-by-step approach. Do it slowly, but do it the right way. Build trust, build confidence,” Jochen Haab, the head of Mercedes-Benz’s autonomous driving program, previously told Drive.
Currently, Mercedes-Benz is rolling out Level 3 autonomous driving technology one market at a time through its Drive Pilot software, which was first introduced in Germany as an optional extra on its S-Class and EQS models from May 2022, followed by the US states of California and Nevada in 2023.
Initially, this allowed the car to be in control up to 60km/h on a freeway only, and the driver had to take back control within 10 seconds. However, the speed was increased to 95km/h in late 2024, and Mercedes hopes it can hit 130km/h by the end of the decade.
Outside of Mercedes and Tesla, brands such as BYD are introducing new-generation ADAS in China – including on more affordable models.
RELATED: BYD launches driver-assistance tech with DeepSeek AI integration
BYD’s ‘God’s Eye’ system has been added to local versions of the Dolphin, Seal, Sealion 6 and Sealion 7 – among other models – but not the Atto 3, as previously reported by Drive.
Also marketed as ‘DiPilot 100’, even the most basic levels of the system include 12 cameras covering the front, side and rear of the vehicle, five millimetre-wave radars, and 12 ultrasonic radars.
‘DiPilot 300′ adds a forward-facing Light Detection and Ranging (LiDAR) sensor and 300 Tera Operations Per Second (TOPS) of peak computing power, while the top-tier ‘DiPilot 600’ adds two additional LiDAR sensors and 600 TOPS of peak computing power.
Some models from Swedish brands Volvo and Polestar have also recently announced an upgrade to gain more capable AI tech.
The 2026 Volvo EX90 and 2026 Polestar 3 have received the new Nvidia Drive AGX Orin supercomputer chip with more processing power, with the 2026 Volvo ES90 also set to get it next.
And even Isuzu has plans to get into the game, with its truck division looking to introduce autonomous vehicles to Japanese roads by 2027.
According to a previous Drive report, the truck brand has not developed its own software but will instead rely on Applied Intuition to add Level 4 automation to its Giga prime movers.
What are the disadvantages of AI in the automotive industry?
As previously mentioned, there are risks when using AI in the automotive industry – unsolicited calls and marketing, substandard repairs, or even motorists believing their vehicles are more capable than they currently are not being responsible.
But one of the biggest concerns is this…
Privacy and cybercrime
Dr Muhammed Esgin, a lecturer at the Faculty of Information Technology at Melbourne’s Monash University, said with any digitisation there is a risk it can be compromised by crooks.
“There have been cases of hacking in these smarter cars. The criminals were able to hack into the car, start the engine and drive it remotely – even though the real driver was not inside,” he told Drive.
“I think the main responsibility [for protecting motorists from car crime] falls on the manufacturers because they’re supposed to put in robust cybersecurity mechanisms in place so that this hacking can’t be done.
RELATED: Elon Musk ‘to ban Apple devices’ from his businesses in row over AI software
“I work on cryptography, and there is really a whole lot of different cryptographic tools you can use to make sure that somebody can’t just easily do this.”
However, with every advancement, Dr Esgin acknowledges, the criminals themselves just become smarter.
When the key first turned from a physical key in the 1990s to a fob that just needs to be near the car to work – criminals started duplicating them.
When car makers introduced ‘digital signatures’ in the 2000s to try and allow just the permitted person(s) to authenticate themselves, thieves hacked this too – even going so far as to chop off fingertips to use on fringerprint scanners.
“As you make systems more complicated, it brings in additional avenues to be exploited by the cyber criminals because every complexity that you add into the system, that opens the door for potential exploitation,” Dr Esgin told us.
“I think we don’t always take security into account when we are designing these systems. I think the main thing is to be perhaps mindful when we are coming up with these new technologies about the security implications.”
Dr Esgin said the same is true of AI and that those designing AI algorithms aren’t always thinking further down the track that it won’t necessarily be used as intended, which is why researchers at Monash are working on more robust cryptographic tools for use within the car industry as well as other digital systems.
“Not everyone is going to use it the same way that the designers intended it for. It’s kind of like kitchen knives – I mean, you can also use them to hurt someone,” he added.
The post How is AI used in cars? The answer might surprise you appeared first on Drive.