The View From DC: Artificial Intelligence

(Part One of the Discussion Paper: The View From DC in November 2023)

Most people are not particularly worried about Artificial Intelligence

AI is clearly disrupting everything – including online safety and I’ve also written about that separately here. In his opening remarks to the conference, FOSI CEO Stephen Balkam noted that AI feels like the third big revolution after the opening up of the world wide web in the late 1990’s and advent of the Web 2.0 in the 2000’s. 

I agree with that summary. Information technology created a step change in the scope, scale and speed of harm incidents when compared to their offline equivalents. Web 2.0 represented another step change – especially in terms of scale. AI has enabled another leap – this time with a particular boost in scope and speed. It is now possible to cause even more harm and disruption, in less time, and with even fewer resources. 

FOSI released this research into the awareness, perceptions, and early use of generative AI tools among parents and teens across the US, Germany and Japan. Respondents expressed safety concerns – especially in relation to disinformation and transparency but generally felt optimistic about the benefits of AI.

AI could be harnessed as a powerful tool in support of positive online experiences. The challenge for the online safety community is now to leverage those opportunities faster than our adversaries. To do that – we’re going to have to combine lots of expertise – and multi-stakeholder collaborations will be more important than ever.

But some experts are (worried about AI)

Some experts have raised concerns about AI and the possibility of “unknown unknowns”. Not only can Generative AI do things better and more quickly, it can also think up new ways of doing things – including bad things.

This is especially front of mind with 64 countries representing 49% of the worlds population heading to the polls in 2024. The ability of AI to create and disseminate disinformation and deepfakes is well understood. These threats have been modelled and systems have been, or are being put in place to combat them.

But AI isn’t just a content generation tool. It is also able to generate new ideas and strategies that could bypass defences, and even worse – go undetected. It is possible that elections will be manipulated, and that the manipulation will not be detected until the election is concluded. Even if that manipulation does not change the outcome of the election, it undermines confidence in the result, and in future elections.  

Even if AI disruption in 2024 is limited to the “known knowns” of AI disinformation and deepfakes – election watchdogs will have a tough year.

Next Section: Online Safety Regulation