WILL ChatGPT SAVE THE CAR INDUSTRY?

“Awful”, this is the usual reaction when using an interface, an interaction in a car. “Great” when testing a NIO or a Tesla. We are in Germany in a car interface crisis. For sure, the traditional car brands are lagging behind. A huge gap many complain about when comparing a smartphone interface with a car screen. And now comes the Large Language Models (LLM), mighty and magic in building conversation with users. New model like LLM will come to the rescue of an industry that has put low emphasis on software, screen and interaction. And always developing on a somewhat “old” model. LLM is the new unstoppable model for interface, even for voice conversation. And voice makes sense in a car.

The gap is growing.

The gap between major technology dynamics and customer interaction in a car is growing. The car industry, and above all the German car industry, has been dominated by mechanical engineering in the last 100 years. In a world where today software engineering reigns and dominates innovation, productivity, growth and customer interaction. An excuse for what we experience in a car - a weird, often non-intuitive interface, a screen that only carries trivial functions like radio and never central functions. An impression that there is no progress at all.

But now comes the Large Language Models like GPT and they are bound to save the industry, catapulting them in new tech and interfaces. Even the economics of a traditional industry under pressure

Under pressure.

The car industry often described as “dominant”is under pressure in its home turf Germany. The arrival of Tesla and a factory near Berlin, the EV growth and - often forgotten - a new generation that has no driving license at all, no car for sure. Under pressure with the troubles at Volkswagen software unit. A unit (Cariad) that was to mark the industry. Under pressure with a screen in a car or an app that takes the main role in new car brands like Tesla or NIO. And consumers that expect a user experience, not an old model of interface and functions from the early days of the internet. Of course, it takes time to design a new car and a new experience. But progress is too slow.

Change interaction to Smart + Voice.

In 2018 we’d expect a big change when Alexa, Google Assistant and Siri grew. Them entering the car because it just made sense: hands on the wheel and voice to control car functions. But it never happened. Car makers developed their own solutions on secondary software (remember “Hey, Mercedes!”?), always avoiding Google, Amazon and Apple. The good idea of “voice in a car” was dead. And training the software was tedious, that is a drawback of solutions based on NLP (Natural Language Processing). Remember the chatbots in 2020!

But still, voice command is such an obvious thing in a car. And training the conversation is not so difficult. Nevertheless, it didn’t work.

Then comes ChatGPT, and a new wave in 2024.

The one-year old ChatGPT has proven to the world how mighty the models can be. The Large Language Models are a game changer. They eliminate the downsides of former models (low memory, finite and hard to train) for an incredible “magic”. A magic everyone experienced asking questions to ChatGPT. Every single professional I know tested it. For the skepticals: the LLM developers continuously work on making them better: faster, less training, less hallucination, get similar answer to same question, sourcing the answer, covering copyright, sovereign, link to recent information and web.

The major investors have decided, this is the new frontier. They are investing billions in the technology - the transformer or foundation model. OpenAI’s competitor Anthropic gathered $7,5 billion within 6 months. X.ai launches its own model Grok. Even German champion Aleph Alpha just gathered $500 million to develop a sovereign version, with Bosch behind.

This is announcing a new wave of LLM for 2024 - all main actors working on a “GPT 5” (GPT4 Turbo being the latest GPT generation). That’s Google’s plan with Project Gemini and Anthropic with the treasure they gathered. It means training and accuracy of LLM will grow covering all questions, ideas, conversations and decisions. They will be fantastically good at conversing with humans, also in cars.

LLM in the heart of a car.

This is a unique chance for the car industry, structurally lagging behind as far as interface, interaction and “experience” is concerned. LLM is the solution, the rescue. Car makers should build all interfaces based on conversation between human and machine, the car. And this is where the LLMs make a huge difference - they are very good at understanding, speaking, translating and training. No more “my NLP isn’t good enough”, the LMM is at the centre and understands what drivers and passengers want.

The solution because all car makers can leapfrog in a new generation of interface model (voice, conversation, long conversation), of software (lighter, open.source often, flexible) and of builders (new generation of developers and designers). Consumers will get used to chatting with a bot, LLM-based. Look at new consumer apps like Pi (Inflection) or Meta AI Assistant with 28 avatar celebrities one can converse with.

Large Language Models belong in the middle of a car, in the middle of a car design and of car maker strategy. We are seeing good first signs of this with Stellantis testing a model on DS (Iris System), NIO with the sweet Nomi steering many functions, Tesla with the sister company X.ai and the latest investment of car supplier Bosch in local LLM developer Aleph Alpha.

More problems solved.

Along with a conversation between driver and car of that quality, new business models will come that have struggled so far. Payment, route planning (Google Maps dominant in car usage), finding petrol stations, operating car functions (secondary ones like light), and any added-service. In fact, it plants intelligence in a car. An intelligence that can be itself autonomous and automatically book parking places, maintenance, EV charging slots. Even help in decision making around such an important object. The car.?

A car saved from convention and habits by LLMs. A whole industry catapulted in the a new technology, in the future.

Hamburg, 23.11.2023

Author: Laurent Burdin is founder of innovation and trend specialist Space and Lemon Innovations in Hamburg/Berlin. He has grown a strong expertise in Artificial Intelligence, coming for math and technology, and landing in business implication. He is writing and speaking in conferences about AI, equipping everyone to understand and accept the dynamic. He also has a long history of consulting for the car industry, with both criticism and optimism.