“We believe that the use and competence of AI among all employees who can use the tools must increase; it shouldn’t be limited to a handful of experts,” says Sven Størmer Thaulow, CDTO at Schibsted.
That sentiment is what lays the foundation for our recent deal with OpenAI – making enterprise accounts for ChatGPT available for over 1,000 employees. The goal is to develop AI skills among employees in a safe and responsible way as the enterprise version provides higher security and privacy, and doesn’t use the input from its users to train OpenAI’s models.
Part of this initiative is also a mandatory “AI as an Assistant” training program which enterprise users within the organisation are required to complete. It offers practical, hands-on instructions tailored to maximise the benefits of generative AI tools. At the end of the training, employees should feel empowered to use AI as a part of their daily work.
The core of the training is making AI work for the user, enhancing their decision-making processes with AI-powered insights, elevating their productivity and efficiency and stimulating their creativity and innovation.
But these are only the latest initiatives taken by Schibsted to increase AI competence in the company. Already in July 2023, we made GitHub Copilot available for all of our software engineers. It’s an LLM trained on programming code, assisting developers by generating code and problem-solving.
“Our developers have been the frontrunners – 60% are using Github Copilot now. With the addition of ChatGPT and other tools, it is time for us to join them and leverage AI to its fullest potential,” says Sven Størmer Thaulow.
With so many of Schibsted’s software engineers using the tool, we have found that it reduces time spent on tedious tasks, allowing them to become significantly more productive and efficient.
In January this year, we also celebrated the 1,000th employee to be enrolled in our AI Academy – an initiative started in October of 2022 with the aim of providing curated training to make AI well-understood at all levels of the organisation. The AI Academy has four tracks:
- The AI Introduction Track, designed for employees who work in non-engineering roles, provides an introduction to AI and how it is used at Schibsted.
- The Deep Learning Track, designed for data engineers, analysts, software engineers, and other technologists with some machine learning experience who are interested in hands-on learning how to train and tune their own AI models.
- The AI Product Track, designed primarily for product managers wanting to increase their understanding of AI and how to identify concrete AI use cases.
- The AI Engineer Track, designed for software developers seeking to learn how to leverage existing AI models.
Still, Schibsted acknowledges that there are some risks tied to the use of AI. We have thorough guidelines detailing the appropriate use of the tool and how to be transparent towards our customers and end-users, as well as advice on the limitations of today’s generative AI:
- Recognise that ChatGPT might not always provide the latest information or context-specific advice.
- Always use your professional judgement to evaluate the advice or information provided.
- Remember to verify critical information through additional sources.
- Stay mindful of biases and cultural limitations in ChatGPT’s training data.