이 웹사이트는 쿠키를 사용합니다. 이 사이트를 이용함으로써 귀하는 쿠키 사용에 동의하게 됩니다. 자세한 내용은 다음을 참조하세요. 개인 정보 정책.

What Can ChatGPT Bring to Smart Cars?

업데이트 시간: Nov 11, 2022 독자층: 3155

ChatGPT's popularity continues, and although it may seem like just a smarter chatbot on the surface, the underlying technology has the potential to expand into various fields and bring about changes across industries. As vehicle intelligence has deepened in recent years, AI technology has been increasingly utilized in automotive applications, such as autonomous driving and intelligent cockpits.

So, what changes can ChatGPT bring to the field of smart cars with this new wave of innovation?

Is it disrupting in-cabin voice interaction experiences?

Microsoft's answer to ChatGPT's practical application is search engines, using conversational interaction to bring disruptive experiences. In cars, a notable trend in intelligent cockpits is to reduce unnecessary physical buttons and shift to touch and voice interactions for controlling different functions.

As an AI model focused on "Chat" functionality, it's easy to imagine ChatGPT being used in voice assistants, including those in car intelligent cockpits that emphasize voice interaction. The voice interaction process involves voice recognition, natural language understanding/processing, generation, and text-to-speech technologies.

Current mainstream providers have achieved high accuracy in voice recognition, and text-to-speech has been widely adopted in various platforms, with AI voiceovers producing highly realistic results.

However, development has stagnated despite voice assistants being widely used in smartphones for over a decade. Voice assistants can often accurately recognize user speech but fail to understand most commands or provide appropriate responses.

car

Today's in-car voice interaction has made some progress in recognizing continuous voice commands, but essentially, voice assistants still recognize only fixed combinations of instructions and perform corresponding fixed actions. As a result, in-car voice assistants cannot be considered truly "intelligent."

The emergence of ChatGPT may bring changes to the intelligent voice industry, as its conversational capabilities have demonstrated the ability to not only "understand" user dialogues but also engage in logical exchanges based on context. Although data source issues may affect content accuracy, combining this type of voice model with in-car applications could significantly improve the user experience of in-car voice interaction.

IFlytek and Cerence currently dominate the in-car voice market; many new car manufacturers have invested in their voice research, such as XPeng and Li Auto, who use iFlytek, Mobvoi, Microsoft, and other suppliers' technologies to develop their distinctive voice systems. For example, XPeng's full-scene voice system can recognize multiple simultaneous commands, differentiate between irrelevant semantics, and allow for instructions while speaking.

If ChatGPT or other similar products enter the in-car voice market, it could significantly impact the technology roadmap of current voice technology suppliers and the overall market landscape.

Recently, Baidu's ChatGPT-like large-scale AI interaction project WenXinYiYan has been announced for integration with Great Wall, Geely, NIO, Airways, and Lantu, but the actual effect will only be known once the product is installed in vehicles.

What does autonomous driving have to do with ChatGPT?

Technically, ChatGPT belongs to the field of natural language processing (NLP), but NLP is not limited to text-based applications. The Transformer model belongs to the NLP field and has been widely used in visual recognition. As a result, ChatGPT may not directly contribute to autonomous driving but may have a greater impact on human-machine interaction. However, the technology behind ChatGPT could drive the development of autonomous driving technology.

In ChatGPT, one difference from previous GPT models is using Reinforcement Learning from Human Feedback (RLHF) to improve the model by judging the quality of its responses based on human-provided content. According to Momenta, this technology can continuously optimize autonomous driving decisions.

공유하다:

이전의:[2023] High Growth of Supercapacitor in Energy Storage and Automotive Applications

다음:What is Energy Harvesting?