Make AI Work
Perhaps you and your team have already tested an AI tool. You might even be using ChatGPT or Copilot on a daily basis or developing AI software yourself. But have you considered what sensitive data you are sharing with AI tools? Do you know the consequences of the European AI Act for your company? Jelle Hoedemaekers, Data Economy Expert at Agoria and speaker at the annual Sirris and Agoria member event ‘AI:Unboxed’, shares some insights and advice.
Jelle Hoedemaekers, Data Economy Expert at Agoria, notes that AI is very well known among companies. ‘A year and a half after the introduction of ChatGPT, many companies are using it. As a result, they have many questions about it, including about the legal and ethical side of AI. We will provide answers and share best practices from companies that have already found their way on ‘AI:Unboxed’ on May 29, 2024.’
In this article you will find insights and advice on three important AI topics:
- Why draw up an internal AI policy
- The positive impact of ethical considerations
- What should already be done about the European AI Act
AI policy describes the use of AI tools
‘We’re seeing that everyone is starting to use AI tools. Often for free. But… nothing is free,’ says Jelle Hoedemaekers. ‘These tools can be fantastic. But they come at a cost and pose risks in terms of data privacy and security. For example, they use your company data to improve themselves. Do you want that?’
This use of data may not be a problem for a quality management system. But when it comes to personal data or trade secrets, it’s a different matter.
Agoria’s advice? Draw up an AI policy. Such an AI policy serves as a guideline for the use of AI within the company, including which tools are allowed, what may and may not be entered and how to handle the results. Make this policy clear. For example, you can use Agoria’s free AI policy template for this. Inform and train your employees based on this policy. See if the Terms & Conditions of AI software are compatible with it. This makes risks manageable.
Ethical considerations on countering prejudice
Discrimination and biases are major challenges of current systems. But you can also look at it differently, according to Hoedemaekers: ‘It is precisely through the use of AI systems that discrimination and prejudices present in the data and the company are exposed. When recruiting software systematically favours male candidates of a certain race or ethnicity, this indicates a historical imbalance in the data and the company.’
The systems themselves do not discriminate, but the data with which they are trained does. But as this discrimination surfaces, it becomes possible to do something about it. ‘You can ask the software to suggest 50/50 men and women or tell it not to take nationality or gender into account.’
Reflecting on this is a first step towards more ethical use of AI. Creating openness in this way and making things discussable is a big and much-needed step.
Jelle: ‘Implementing ethical AI systems is often seen as a hurdle, but it actually offers many benefits. Companies that integrate ethics into their AI projects from the start find that the risk of failure is reduced. Moreover, it promotes innovation. Ethical considerations force them to think more thoroughly about the applications of AI and its impact on people. This leads to more thoughtful, inclusive and ultimately successful solutions.’
Agoria’s advice? Implement an ethical framework. This does not have to be built from scratch. ‘Unfortunately, there is a proliferation of ethical frameworks and a one-size-fits-all does not exist. Which solution suits your company depends on its size, activity, sector, etc. The implementation will also be different in a company of two or two hundred people.’
No one can escape the European AI Act
The European Union is about to introduce new legislation specifically aimed at AI. This is an important moment for all companies implementing or using AI. Developers and users will have additional responsibilities. ‘Companies are wondering what they need to do to make their products compliant and which standards they need to integrate,’ says Jelle Hoedemaekers. ‘It is already clear that this legislation will lead to more rigorous compliance processes. This is not obvious especially for smaller companies, startups and scale-ups.’
Agoria’s advice? Start considering how the European AI legislation might affect your business today. You still have two years to make products and use compliant. But don’t wait for this deadline. Start educating yourself and your legal team. Look for companies you can learn from that are already tackling this challenge. Find answers in knowledge that is shared online and at events.’
AI:Unboxed - Make AI Work!Would you like to get started with Artificial Intelligence (AI) yourself or gain insights into what this technological revolution could mean for your company, your job and all companies in the digital and manufacturing industry? Come to AI:Unboxed, the annual members event of Sirris and Agoria at the Brussels Expo on 29 May 2024. The programme includes, among others, a keynote by Prof. Ethan Mollick, a panel discussion about the European AI ecosystem, several (fully booked!) breakout sessions and a Tech Fair with spectacular real AI applications. In his session, Jelle Hoedemaekers will discuss the European AI Act, ethics as a solution and the implementation of an internal AI policy. |
