Superbowl LXV (so, 2031), will be contested between the Tennessee Titans and the Arizona Cardinals, with the AFC Champions coming out 17-15 on top with a walk off field goal from 51 yards.
That’s about as probable a prediction as any (although, based on historical performances, you’d be forgiven for suggesting it might err on the improbable).
Why the reference to sports? Well, it’s to illustrate that any prediction of the future will be precisely that – a prediction.
Regular readers of this blog will know my background in the dark compliance arts and those that pay close attention will know the only guarantee I give about my predictions is that they will be wrong. It’s just a question of how wrong.
I am pretty confident that governments will seek to regulate AI (once they figure out what it is, and truly distinguish AI from just brute force computing), just to be seen to be “doing something”. I will also predict that, absent directly addressing something that offends their flagship policies, such as Elon Musk recently found out in the UK, Western governments will take a light touch.
The balance for our parliaments will invariably involve allowing AI to address the productivity problems that have been endemic in many economies for decades, and concerns about lost jobs.
So, we will likely have a wait and see approach, and the market will be allowed to develop for now. After all, we already have a plethora of consumer protection laws, cybersecurity laws, privacy laws and a regulatory framework for technology generally. Case law is developing about the liability for bad AI decisions (quite how Air Canada’s General Counsel ever allowed this case to see the light of day is beyond me – and UnitedHealth is at the forefront of issues in that sector) but those cases are following traditional patterns of tort. Your agent (human or machine) went and messed up, here’s your penance.
The EU led the way at the start with its AI Act, but, if you know your EU laws and GDPR, you’ll see a lot of similarities in the underlying structure and will of Parliament across them and there’s nothing truly radical there other than the EU managed to codify common sense for a change.
I also predict (and this is a thread that has come out of the excellent Cavell Group conferences) that humans will place an increasing premium on human interaction. Ryanair will undoubtedly let an AI agent tell you that it doesn’t care about your lost bag any more than the laid-off human call centre agent did. High-end brands, like Emirates, may have “No AI here guarantee” on their contact centre. The market will differentiate itself as it always has, and consumer economic pressures will work to keep a degree of restraint on many CFOs’ desires to dispose of wage bills.
But beyond that? Any more detailed prediction is a finger in the air. Ultimately, our legislators don’t know what they are going to be legislating for or against. Anyone out there preaching that this government will enact that law, and this regulator will pursue this rule, needs to be treated with the same pinch of salt as my Superbowl prediction deserves.
Development cycles in this field are working in weeks and months; governments work in years and any good points made in a Parliamentary debate will have an increasingly short half-life. We’ve seen this ourselves already – the difference between v1.0 and v2.0 of our own Conservational AI (across 8 months) is immense, and the roadmap for the next drop in a couple months is another order of magnitude improvement.
But I can point to one thing, and this is something that should encourage you to start your AI journey with Simwood with some confidence.
The UK (where Simwood is ultimately headquartered and owned) has one of the most stringent telecommunications security laws in the world. Despite Brexit, it still retains the EU’s GDPR regime, which is one of the most stringent privacy laws in the world. Our Conversational AI has been built assuming that governments are going to meddle in ways that will make no sense.
We’ve lived through rules on call recording, PCI compliance, GDPR, debates on interception laws, the move to over-the-top VoIP from TDM and more.
So when you’re thinking about where to place orders to satisfy your big blue chip enterprise customer’s AI aspirations, there’s a lot to be said for doing it with a company that knows the Sheriff will probably visit the Wild West sooner or later.
