Artificial intelligence (AI) is a subject that sparks divergent opinions among experts, who generally fall into one of two camps: those who believe it will significantly enhance our lives and those who fear it may lead to our demise.
This is why the recent debate in the European Parliament regarding the regulation of AI holds great significance. The crucial question at hand is how to ensure the safety of AI technology. Below are five key challenges that lie ahead in this regard.
Establishing a common understanding of AI
After a two-year-long process, the European Parliament has finally crafted a definition for an AI system. According to their definition, software that can "for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations or decisions influencing the environments they interact with."
This week, it is voting on its Artificial Intelligence Act - the first legal rules of their kind on AI, which go beyond voluntary codes and require companies to comply.
Achieving global consensus
Former head of the UK Office for Artificial Intelligence, Sana Kharaghani, emphasizes that AI technology knows no boundaries and requires international collaboration.
"We do need to have international collaboration on this - I know it will be hard," she tells BBC News. "This is not a domestic matter. These technologies don't sit within the boundaries of one country
Different regions have their own perspectives on AI regulation:
- The European Union has proposed stringent regulations, including the classification of AI products based on their impact. For instance, a simple email spam filter would face lighter regulation compared to a cancer-detection tool.
- The United Kingdom is integrating AI regulation within existing regulatory frameworks. In cases of AI discrimination, individuals could approach the Equalities Commission.
- The United States has primarily relied on voluntary codes, but concerns have been raised about the effectiveness of such measures, as acknowledged in a recent AI committee hearing.
- China intends to require companies to notify users whenever an AI algorithm is employed.
- Balancing innovation and safety: Striking a balance between encouraging innovation and ensuring the safety of AI poses a significant challenge. Regulators must avoid stifling progress while addressing potential risks associated with AI deployment.
Accountability and transparency
"If people trust it, then they'll use it," International Business Machines (IBM) Corporation EU government and regulatory affairs head Jean-Marc Leclerc says.
Holding AI developers and users accountable for the actions and decisions made by AI systems remains a crucial challenge. The ability to trace and explain the reasoning behind AI-generated outputs is essential for building trust and ensuring ethical practices.
Addressing biases and discrimination: AI systems have shown the potential to perpetuate biases and discriminate against certain individuals or groups. Overcoming this challenge involves developing mechanisms to detect and mitigate bias in AI algorithms, as well as establishing guidelines to ensure fair and unbiased use of AI technology.
As the European Parliament prepares to vote on the Artificial Intelligence Act, the first set of legal rules specifically designed for AI, the outcome will significantly shape the future of AI regulation in Europe and possibly serve as a reference point for global discussions on the subject.
Determining Responsibility for Rule-Writing
Up until now, the regulation of AI has primarily been left to the AI industry itself. While major corporations claim to support government oversight to mitigate potential risks, concerns arise regarding their priorities. Will profit take precedence over people's well-being if they are heavily involved in shaping the regulations?
It's highly likely that these companies want to exert influence on the lawmakers responsible for establishing the rules. And Lastminute.com founder Baroness Lane-Fox says it is crucial to listen not just to corporations. We must involve civil society, academia, people who are affected by these different models and transformations," she says.
Acting quickly:
Microsoft, which has made significant investments in ChatGPT, aims to streamline work processes by eliminating tedious tasks. While ChatGPT can generate text that closely resembles human writing, OpenAI CEO Sam Altman emphasizes that it is a tool, not an autonomous entity.
Chatbots are designed to enhance productivity for workers and can serve as valuable assistants in various industries. However, some sectors have already experienced job losses due to AI implementation. BT, for instance, recently announced that 10,000 jobs would be replaced by AI.
ChatGPT became publicly available just over six months ago and has rapidly expanded its capabilities. It can now write essays, plan vacations, and even help individuals prepare for professional exams. The progress of these large-scale language models has been astonishing, prompting warnings from renowned figures like Geoffrey Hinton and Prof Yoshua Bengio, considered the "godfathers" of AI, about the immense potential for harm.
The implementation of the Artificial Intelligence Act is not expected until at least 2025, which EU technology chief Margrethe Vestager deems far too delayed. In response, she is collaborating with the United States to develop an interim voluntary code for the AI sector, with the possibility of completion within weeks.
