The risks and drawbacks of AI
Like any powerful technology, AI brings both opportunities and challenges. While much has been said about its potential, it's just as important to understand the downsides: the risks, blind spots and unintended effects. We take a look at the foreseeable drawbacks of AI, as far as we can tell in a field that’s still evolving at lightning speed. Understanding the technology will always be the first step towards mitigating risks: to stay in control, we need to keep learning.
Costs, accessibility and ecological footprint
- Developing and using AI requires significant computing power, leading to high energy consumption and increased carbon emissions. (What is the environmental impact of AI)
- To train AI systems, masses of data need to be labelled first. Then the system must be intensively tested. These tasks are performed by people, often in poor conditions in low-wage countries.
- Because AI system development is so expensive, small businesses and developing countries may struggle to keep up with rapid technological advancement.
Dependency and control
There is a risk of AI systems operating without sufficient human oversight, leading to unforeseen and potentially harmful consequences such as incorrect treatment of diseases, wrongful convictions, discrimination, and even killings by autonomous robots with facial recognition.
- Many educators worry that students will no longer master basic skills ⇗ if they rely too much on AI systems.
- So called AI therapists are proving to be both beneficial and high-risk
Employment and professions
- AI automation can lead to job loss in some sectors. Which sectors could be affected by AI? AI affects traditionally ‘human’ sectors that long seemed impervious to automation, such as translation, journalism, software development, graphic design, and even text and music composition.
- People may even become redundant, for example, actors being replaced by lifelike animated people and synthetic voices.
- People need to retrain and adapt to new technologies, which requires time and resources.
Complexity and transparency
- Many AI models, particularly neural networks, are complex and difficult to understand (their internal processes are like a black box).
- Lack of transparency can lead to distrust and difficulties in justifying decisions.
- AI systems are statistical models that make decisions based on probability. Unlike traditional computer systems, they are not ‘always right’ and should not be blindly trusted.
Our tips to start learning about AI:
Privacy, security, and copyright
- AI systems can collect and analyse vast amounts of personal data, leading to potential privacy breaches.
- Sensitive data can be vulnerable to hacking and abuse.
- AI systems can replicate the work of authors, artists, and content creators, potentially violating copyright laws.
- Criminals may use AI systems like deep fakes for phishing and blackmail to deceive people more convincingly. This may include the creation of realistic voices, photos, and videos.
Bias and discrimination
- AI recognises and reproduces existing patterns. As a result, existing biases and discrimination in data are identified, reinforced, and reproduced.
- Lack of diversity in data can lead to a lack of representativeness.
Ethics and responsibility
- There are ethical issues surrounding the decisions made by AI systems, for example, in healthcare and law.
- It is often unclear who is responsible for the actions and decisions of AI systems, for example, for self-driving cars, but also for diagnoses or subsidies.
- AI can fabricate information that is inaccurate or untrue. Even references are sometimes ‘hallucinated’. Checking with a reliable source is always necessary.
How to protect yourself from the drawbacks of AI?
The AI train has left the station, and there’s no stopping it now. But let’s not forget: AI is just another technology. Like any tool, it can be used for good and bad.
By learning more about AI and gaining a clear view of the real risks, you can help steer its direction in your own profession. Here are some key recommendations:
-
Know what’s going on. Make sure your company or management knows which AI systems are being used in your team. Agree on a clear code of conduct and ensure human oversight for higher-risk systems. AI literacy (EU AI Act Article 4)
Familiarize yourself with the EU AI Act: it already mitigates a lot of risks and even outlaws those AI systems that have an "unacceptable" risk. Start f.i. with our blog What is the AI Act?
-
Make learning and innovation part of the company culture. Build a learning mindset in your organisation, and don't forget that informal learning matters too. A shared knowledge base helps to keep know-how accessible and unties it from individual team members (Read more in How to increase AI Literacy in your organisation? Answer: training programmes and a learning culture)
-
Play your part. Whether you’re a user, end user, or manager - you have a role in shaping new (AI) systems. AI affects the processes that run your organisation, so it shouldn't be left to IT only. (See Waarom leren en kennis noodzakelijk zijn om AI-systemen te mogen vertrouwen)
-
Look for hidden risks. New developments always carry risks. But there's no need to dive in unprepared: tools like the AI blindspots card set of Knowledge Centre Data & Society help you to properly assess risks and to include different perspectives.
-
Be critical and constructive. AI won’t replace your real-world expertise - unless you stop developing it Speak up, keep learning, think along, and use AI to take over the tasks you don’t enjoy. (Read The friction between learning and doing: am I not better than AI?
Learn more
The friction between learning and doing: am I not better than AI?
Frederik Picard confesses: he is addicted to beating the AI machine. Despite AI clearly being of added value to his professional life, he still feels the urge to outperform ChatGPT. This competition serves not only as motivation to improve his abilities but also as a reminder of the unique power of human creativity and ingenuity in a world where technology is becoming increasingly dominant.
How do you recognise an AI text?
Bavo Van Landeghem of Scriptorij explains how to recognise AI texts and gives tips on how to use AI tools as writing aids.
Isabelle Borremans
Isabelle is not an AI expert, but she has been communicating about AI (and AI training) for VAIA for four years. So, she knows exactly how to communicate effectively about AI training. Feel free to contact her with any questions on how to promote your AI course. She’ll gladly challenge you with questions like:
- Who is your target audience? Can you be more specific?
- What networks reach that audience?
- Where does your audience prefer to take courses?
- What AI skills does your audience need?
- Does your training match their needs?
Besides communication, Isabelle also works on VAIA’s strategy: Who is our audience? How can we reach them? How can we spark their interest in AI?
Related courses
Machine learning and deep learning fundamentals
Workshop - Leuven - VIB, Training geselecteerd in de Belgische AI Factory Antenna (AIFA)
European Agentic AI
Bootcamp - Online - lrz, Training geselecteerd in de Belgische AI Factory Antenna (AIFA)