Welcome to 2024: Volkswagen really is putting ChatGPT into cars as a gabby copilot
Plus: Deloitte rolls out homegrown AI for its 75K workers to dogfood
CES Volkswagen says it intends to inject ChatGPT into its vehicles in Europe this year, allowing drivers to ask questions about life, the universe, and everything, and have them answered by the software, maybe.
Announced to coincide with the CES show in Las Vegas this week, the German automaker said the OpenAI chatbot will be accessible via VW's Ida voice-controlled assistant, which launched in 2023.
Ida is bundled with Volkswagen's ID.7, ID.4, ID.5, and ID.3 all-electric rides, plus its latest Tiguan, Passat, and Golf models. If you have a vehicle with Ida, it'll automatically be able to use ChatGPT, we're told, with no need to open an account, install an app, or, er, opt-in. Prior to Ida's arrival, the automaker had been adding voice-based controls to its cars; Ida seemingly grew from that, and now has ChatGPT plugged in.
Ida can already answer some spoken natural-language questions about the weather, stock prices, sporting events, and the like – just what you want to know while driving around town. It is activated by pressing a button on the steering wheel or by waking it up with: "Hello Ida."
Some reviewers previously warned not to say the word "idea" inside your VW, or you might accidentally stir the bot from its slumber. You can also ask the software to adjust the temperature in the car, change whatever the infotainment system is doing, set a route, and so on.
What's new here is that Volkswagen has tasked Cerence – a startup that emerged in 2019 from the now-Microsoft-owned Nuance – to integrate ChatGPT into Ida, allowing the bot to be accessed via voice while driving.
We're told that when you wake up Ida and talk to it, it will figure out whether you're asking it to perform an action specific to your vehicle – such as adjusting the aircon, or looking up a destination. If it can handle the job itself, it'll do so. If not, it will reach out to Cerence servers, which may in future ask ChatGPT to solve the problem and generate an answer.
According to VW's marketing:
ChatGPT does not gain access to any vehicle data; questions and answers are deleted immediately to ensure the highest possible level of data protection. This is facilitated by Cerence Chat Pro, which leverages a multitude of sources, including ChatGPT, to enable IDA to provide accurate and relevant responses to nearly every query imaginable. The feature also prioritises security and seamless integration with IDA’s myriad capabilities, delivering ease of use for drivers.
So Volkswagon is at least familiar with the concept of privacy and security. On the one hand, the hands-free, information searching-slash-guessing code, could be useful while driving or trying to settle some argument about trivia, depending on how accurate the AI is or isn't. It may stop some people fiddling with their phones while behind the wheel. On the other hand, well, how much do you want to trust this stuff?
Knight Rider's KITT, it is not.
Volkswagen said it will be "the first volume manufacturer to offer ChatGPT as a standard feature" when that arrives in the second quarter of 2024 in Europe. The manufacturer may roll it out further around the globe later on.
Speaking of new AI...
Consulting giant Deloitte has powered up an internal large language model for its 75,000 employees to write emails, design PowerPoint presentations, and generate code, and to help iron out any bugs in the system.
The tool, dubbed PairD, was developed in-house by Deloitte's AI Institute and will be rolled out to staff working across Europe and the Middle East, as well as 800 employees at Scope, a UK disability equality charity. Deloitte will train its staff on how to use the chatbot to replace help them carry out their duties.
"Generative AI should be available and accessible to everyone and businesses like ours must ensure that the adoption of AI promotes social equality rather than exacerbating existing disparities," Richard Houston, CEO of Deloitte UK and Deloitte NSE, said Monday.
"But it's not only about access to technology – it's access to the skills to use it that will really create opportunity to help close the digital divide. We want to increase the accessibility of our AI platform."
Like all LLMs, however, PairD isn't perfect and can generate false information. Employees were warned that the system could make up facts, and were advised to double check the chatbot's responses for "accuracy and completeness," the Financial Times noted.
- Here's who thinks AI chatbots will eventually be smart enough to be your coworker
- Everyone's talking about AI but industry reps say few are ready to implement
- Boffins find asking ChatGPT to repeat key words can expose its training data
Last year Deloitte's rival PwC began testing ChatPwC, an internal tool powered by OpenAI's tech, for its own employees. The system reportedly runs in a custom-built secure Microsoft Azure environment to prevent the accountancy titan's private data from leaking out one way or another.
In October, PwC also announced that it was collaborating with OpenAI and Harvey, another LLM developer, to train more large language models on its own data to automate tax, legal, and HR services. Bivek Sharma, global head of AI for tax, legal and people, PwC United Kingdom said generative AI "will be the biggest game-changer for advisory services for a generation."
Organizations buying into this AI hype will need to take steps to keep a lid on any proprietary information being processed by models hosted in a public or semi-public cloud. Apple, JPMorgan Chase, Amazon, Samsung, and others have previously banned staff from using ChatGPT and similar remote services over fears that private data and trade secrets could be inadvertently and inappropriately disclosed beyond their corporate borders. ®