Are we doing enough to regulate AI?

The EU has taken the lead in developing a regulatory framework for AI according to the FT.

Although lagging behind the US and China in terms of its regions actual AI capabilities, the EU has taken the first major steps towards developing a regulatory framework for AI that will foster long-term consumer confidence in the industry.

Two areas stand out relating to the EU’s recent flurry of tech related regulation.

Firstly, the GDPR (General Data Protection Regulation) has a number of implications for AI. It includes a number of safeguards to prevent automatic decision-making in the absence of human involvement. Such decision making is only allowed if certain preconditions are met. For example, an individual who applies for a loan online and is given an instant ‘outcome’ could be entitled to receive meaningful information about the logic behind that decision. In effect, this measure would allow stakeholders to scrutinise these companies more hopefully ensuring they were using AI appropriately.

This evidently has strong implications for how companies use AI, despite the GDPRs claims that the legislation is ‘technology-neutral’- an entirely fallacious description.

Secondly, the EU soon plans to launch its AI code of conduct, a relatively unambitious attempt at tech ‘self-regulation’. This will include a number of proposals including concrete measures that encourage sound system design and provisions that these developments ‘increase human wellbeing’. Whilst certainly admirable in spirit, it is unclear how these provisions will actually regulate companies given that these measures are not legally binding and will remain voluntary.

Small steps definitely, but there is something to be said for embracing this as a 'step in the right direction'. Firstly, this will encourage other governments to think about how they regulate AI- expect much cross-national policy transfer and learning to occur in this space. Secondly, EU regulatory policy happens incrementally; in establishing ‘policy credibility’ here, the EU is demonstrating its ability and willingness to introduce more radical steps in the future; from asylum policy to the liberalisation of voice telephony , this is typically how things go down in Europe: small steps and then bigger ones.

However, as noted earlier, the EU is a capable but crucially not leading figure in AI development- China and the US share this mantle. Perhaps these great powers will take note, but will they want to risk tying up industry with red tape, potentially at the risk of the other nation gaining at their expense?

Time will tell on this front, but it is highly likely these governments will prioritise development over regulation, in what could potentially mirror an ‘AI race’.

Evidently there is much to be concerned about. Will the EU do enough? Will China and the US bother?

The answer to these two questions could prove to be crucial on so many levels, let’s hope that all parties involve make the right decisions.