Brakes on a Rocketship?
With new generative AI systems and applications being released, announced or demonstrated daily, it is hardly surprising that this new breed of tools is also catching the eye of regulators and law makers around the world.
Chat GPT and its (competitor) relations are providing an increasingly clear vision of what lies ahead. The combination of machine learning algorithms, previously unimaginable compute power and a learning “database” that comprises the entire internet means that all kinds of efforts that would have previously been impossible, are within reach. Companies, startups, governments and universities the world over are rushing to explore what’s possible today, and what might be possible next.
While today’s tools can surprise us with their capabilities, they can also shock with their capacity to dissemble, or technically — to “hallucinate”. More plainly: they lie, and lie with conviction!
Two weeks ago an open letter was signed by some 25,000 people {Ed — well and probably some robots}, including Elon Musk, calling for a moratorium on the work of AI labs to train tools more powerful than GPT-4 for at least 6 months.
This week a key US agency is seeking comments on whether and if so, how, AI tools should be audited or licensed. The EU and the UK have been focussed on Ethics in AI for some time.
Is this an area where regulation can move as quickly as technology? Should it? Has the rocket already reached escape velocity?
While we can, and we should, be concerned with the manner in which these tools develop and the way that their use can be constrained for positive use, it would be foolish not to be thinking about ways that advanced AI tools will impact our efforts in digital reporting. Our own view is that many aspects of these technologies will be profoundly important for XBRL — and in a positive way! A new area of focus for this newsletter, as well as for many of you. Stay tuned!
