What is ChatGPT?
ChatGPT is an AI-powered large language model (LLM) built on a sophisticated neural network developed by Microsoft-backed startup OpenAI. GPT stands for Generative Pre-trained Transformer, a family of LLMs trained to create or summarize texts, answer questions, translate languages and even write programming code. The dialog format makes it possible for ChatGPT to reply to follow-up questions, admit its mistakes, challenge incorrect premises and reject inappropriate requests.
How is ChatGPT different from traditional chatbots?
- ChatGPT allows users to ask questions in natural language, making interactions more intuitive and user-friendly. Its advanced language model enables it to comprehend and respond to complex queries that may be challenging for traditional chatbot systems.
- ChatGPT can understand context, learn continuously from new data and adapt its feedback accordingly. The model can memorize previous messages in a conversation, which permits it to generate remarks that are relevant and makes sense in the context of the conversation. This is a huge advantage over traditional rule-based chatbots, which rely on preprogrammed statements and can often give irrelevant or nonsensical answers.
Potential ChatGPT automotive use cases
- Enhancing customer service. One of the most immediate applications of ChatGPT in the automotive industry is customer service. Companies can integrate chatbots into their websites or mobile apps to benefit customers with instant help and support. ChatGPT can respond with useful information or direct customers to the appropriate department for further assistance and reduce their wait times. ChatGPT can also prove valuable for automotive dealerships looking to improve their customer engagement and lead generation.
- Increasing vehicle autonomy. Merging ChatGPT with autonomous vehicles can potentially fast-track their decision-making capabilities for more informed and intelligent decisions. Automakers can assimilate chatbots into the car's interface to provide passengers with knowledge about the car's status and upcoming actions; for instance, taking alternate routes to avoid traffic congestion or identifying the nearest paid parking spot near a restaurant. This incorporation may also facilitate vehicles to swiftly anticipate and appropriately respond to risky scenarios, such as avoiding a jaywalker.
- Supporting sales and marketing. By leveraging ChatGPT, car dealerships and manufacturers can create conversational sales assistants that guide customers through the buying process. Chatbots can form specialized recommendations based on a customer's preferences and needs, making the buying procedure more efficient and convenient. Companies can also utilize ChatGPT to tailor high-quality content for marketing purposes, such as eye-catching brochures, informative blog posts and compelling news articles.
- Improving manufacturing processes. Original equipment manufacturers and suppliers are using the technology to produce digital assistants that can help workers on assembly lines. These assistants can enact step-by-step instructions for assembling different components, reducing errors and downtime, and maximizing efficiency and productivity.
- Supply chain management. By analyzing data from suppliers and logistics partners, ChatGPT can support OEMs to optimize their supply chain and reduce costs. Chatbots can also prepare real-time updates on status of shipments and help manage inventory.
- Repair and maintenance. By combining ChatGPT into the car’s onboard computer, the chatbot can interpret diagnostic data and give recommendations for repairs or maintenance, thus minimizing the likelihood of breakdowns.
From an end-user perspective, some of ChatGPT’s in-vehicle use cases include:
- Creating personalized voice assistants that can interact with the driver in a natural way and presenting information or suggestions based on the driver’s preferences.
- Providing car health and smart car manuals that eliminate the need for printed manuals.
- Generating a voice-controlled navigation system that allows drivers to input destination and other commands using natural language — easily and without the need to take their eyes off the road.
- Analyzing a driver’s past navigation history and predicting their likely destination based on their usual habits, thus helping drivers to save time by automatically contributing directions to their most frequently visited locations.
- Evaluating real-time traffic data and communicating updates on traffic conditions and alternative routes, thus enabling drivers to avoid delays.
- Introducing customized recommendations to drivers for points of interest, such as restaurants or gas stations, along their route.
What are the limitations of ChatGPT?
- ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers.
- ChatGPT is sensitive to tweaks to input phrasing or attempting the same prompt multiple times.
- ChatGPT is often excessively verbose and overuses certain phrases, such as restating that it is a language model trained by OpenAI.
- ChatGPT sometimes responds to harmful instructions or exhibits biased behavior.
What are the potential challenges in using ChatGPT?
- One of the main concerns of using ChatGPT in the automotive industry is the risk of data privacy and security breaches of sensitive information, such as the driver's personal data.
- While ChatGPT is capable of understanding context to some extent, there may still be limitations in its ability to fully understand the context of a conversation or command.
- ChatGPT can be non-feasible for smaller automakers as it is trained on very large amounts of data.
- As with any technology that can generate human-like text, there are ethical concerns about the potential misuse of ChatGPT to create deepfake videos and images, impersonate real people online and spread misinformation.
Which automotive OEMs are working on ChatGPT?
Mercedes‑Benz: Mercedes‑Benz has a collaboration with Microsoft to enable integration of ChatGPT through the Azure OpenAI Service. The automaker apparently started an optional over-the-air beta program in the US in June for over 900,000 vehicles equipped with the MBUX infotainment system. By adding ChatGPT, the MBUX’s voice-control capabilities will become even more intuitive. Mercedes-Benz plans to use the findings from the beta program to further improve its voice assistant and define the rollout strategy for LLMs in more markets and languages. The automaker stores the collected voice-command data in the Mercedes-Benz Intelligent Cloud, where it is anonymized and analyzed.
BMW: BMW has incorporated ChatGPT into its iDrive system, letting drivers control various functions of their car using natural-language voice commands. This includes features such as adjusting the temperature, changing the radio station, and even opening and closing the sunroof.
Ford: Ford has also implemented ChatGPT in its in-car entertainment and navigation system. It allows drivers to control the system using natural language, and supports real-time traffic updates and personalized recommendations for points of interest.
General Motors (GM): GM is likely exploring uses for ChatGPT as part of its broader collaboration with Microsoft. GM and Microsoft have been working together on driverless vehicle technology since 2021, when Microsoft invested in GM’s autonomous robo-taxi service Cruise. GM Vice President Scott Miller had said earlier in 2023 that GM could possibly use ChatGPT to access material on how-to-use vehicle features normally found in an owner’s manual, program functions such as a garage door code, or including schedules from a calendar.
Toyota: While there is no specific news available regarding Toyota's plans to develop its own AI-powered virtual assistant similar to ChatGPT, the Japanese automaker has been actively exploring the fusion of AI and voice-recognition technologies in its vehicles. Toyota is apparently phasing out support for Amazon’s Alexa in its vehicles as it focuses on improving its in-house voice assistant, likely by assimilating ChatGPT.
Mainland China: As with most new automotive technologies, OEMs in mainland China are already working extensively with LLM-based AI chatbots. The likes of SAIC, Great Wall Motor, Nio, Li Auto, Xpeng, Chery and Geely are working to implement ChatGPT-like capabilities in their new vehicles, to enhance their voice assistants for free chat and storytelling, location-based tourist spot recommendations, AI-based painting, multimodal in-cabin sensing and car manual instructions, among other features.
At the 2023 Shanghai Auto Show, there were a plethora of new AI-based LLM launches, including:
- Banma Co-Pilot, the third-generation, car-AI capability system of Banma Zhixing. Based on the Tongyi Qianwen large-scale model, it is meant for automotive scenarios and builds a cloud-integrated full-stack AI capability. Its core capabilities include scene customization, multisource data, professional knowledge, service access, behavior prediction, etc.
- SenseTime’s SenseNova large-scale model system, including application platform Consultation SenseChat, image-generation model Miaohua SenseMirage, AI-digital-human video generation platform SenseAvatar and 3D-content generation platform Qiongyu SenseSpace, Gewu SenseThings, etc.
- iFlytek: The company is becoming a reckoning force in this domain, with multilingual project cooperation with SAIC, Changan, Chery and many others, covering dozens of languages such as English, Japanese, Thai, Spanish and Italian. In May, the company launched Xunfei Xinghuo Cognitive Model + Smart Cockpit, which can provide human-computer interaction paradigms for multiwheel, multiperson, multiregion and multimodal smart cars.
- Baidu: At the time of writing, Baidu Apollo announced that Great Wall Motor and ECarx became the first batch of application exploration partners of Baidu’s Wenxin large-scale smart cabin — also known as Ernie Bot — and touted as ChatGPT’s mainland Chinese rival. Baidu Apollo has completed verification of many of Ernie Bot’s interactive functions on mass-produced vehicle platforms, including Great Wall, Lynk & Co., and Smart.
Conclusion
LLMs, like the one powering ChatGPT, have shifted the AI-development paradigm with their ability to understand context and generate nuanced responses. While ChatGPT is a powerful language model that has the potential to greatly improve the driving experience, it is important to be aware of the potential challenges and limitations of using it in an automotive environment. Stakeholders need to carefully consider and implement proper measures to ensure data security and privacy, and address ethical and interpretability concerns.
Despite these risks, the integration of ChatGPT in the automobile industry is becoming inevitable as the use cases are compelling. Given the complexities involved, it will require a significant amount of time to fully leverage the automotive-related benefits of ChatGPT and similar models. Currently, ChatGPT is probably the most popular AI chatbot but researchers are actively testing others, including those targeted specifically for automotive uses cases.
S&P Global Mobility believes that going forward, OEMs will have to work very closely with their chosen AI technology provider to not only upgrade the process of data collection from the vehicle, but also set up advanced LLM trainings to enhance overall AI capability, and provide more comprehensive services and functionalities.
