This is a thought experiment, albeit one that I believe will be less an experiment and more an experience in the not too distant future.
In one way or another, beneficial or insidious, AI augmented systems are beginning to affect our lives. Our world now has instant language translation, auto-bot DJs that know our musical tastes, autonomous vehicles, and facial recognition using AI.
The thing is that the majority of these systems are “big AI”. There is personalisation involved (such as my musical tastes, my location, articles I click on…), but it’s not like it’s my AI, it’s their AI.
I think this will change over the next few years as we find AI increasingly able to be personalised to our preferences and to act for us as individuals, not them.
Here’s an example prediction.
None of us really read the Terms and Conditions of the various online services that we sign up to. It does not matter that the User Interface makes you scroll down the page before you can hit “accept”, you don’t actually read the stuff. Yet you know probably should, even if (depending on which country you are reading this in) you have some additional protection under common and/or consumer law. Why? For in those reams of words are important statements on how the service provider will treat your data, what liabilities they do and don’t accept etc, and which leg you can or can’t stand on.
In their oft cited 2008 paper “The Cost of Reading Privacy Policies” by Aleecia M. McDonald and Lorrie Faith Cranor, published in the I/S: A Journal Of Law And Policy for the Information Society, the authors provide a point estimate that 244 hours would be required every year to read the privacy policies presented to web users (at the time apps were next to non-existent, hence this focus).
Let’s use an 8 hour working day as a crude measure, hence 244 hours is about 30 days lost every year just to reading the policies. Using 2008 labour cost rates, for the US alone, they estimated a cost of $781 billion US per year! As someone who used online services in 2008, and still using online services in 2018, the last ten years don’t seem to have made things better.
So we have a problem that seems worth solving for a number of consumers:
1. Companies that offer privacy policies and other Terms and Conditions that don’t actually want them to be “wrong”, to have inconsistencies, or to be unenforceable.
This is a benign view I know. These companies want to ensure that their T&Cs are correct and comply with local legislation. They also want their maximum enforceable protection, but it will actually cost them effort and possibly fines if they have incorrect policies and agreements. Compliance isn’t easy…
2. People reading privacy policies and Terms and Conditions who should be aware of what they are signing up to.
At the very least these documents describe how the service provider intends to behave, even if that behaviour isn’t actually legal locally. In this circumstance, no-one but the lawyers win if you have to fight the contract, so it is best to recognise the risk upfront and avoid the issue if you can.
At some point in time, someone is likely to have to arbitrate over whether a policy is legal or not in whatever jurisdiction the argument is being held. This process is generally time consuming and messy for everyone (except the lawyers).
Right, so it’s risky for all parties, time consuming and hence expensive even just to read the things. Oh and Terms and Conditions change – we didn’t read them the first time around, so we’re even less likely to read them when they’re updated.
Enter My Little AITorney (I’d try to trademark this and have a logo with a mane, but I think I’d be sailing too close to some large commercial interests…)
My Little AITorney is personalised AI in that you tell it what your preferences are with respect to sharing information in general, and it learns as you act – advising you if it thinks it needs to adjust parameters based on observed behaviour. Critically, it reads the contracts before you sign-up and warns if something is wrong for your preferences (or local laws).
Now I know there is already a growing arsenal of legal AI systems reading commercial contracts for law firms. However, I’m talking personal consumer level here, “prosumer” if you will. This has to be something that you spend pocket money sums on, not hundreds or thousands of dollars in seven minute increments.
So My Little AITorney is watching my back, or sitting on my shoulder depending on your visualisation. It reads the contracts of services that I am about to sign up to, and advises if there is anything I should know. It re-reads any advertised changes and advises me of them and whether there is anything new I should know. It understands relevant background local and global legal frameworks such as consumer protection law and advises me if there is a conflict. Perhaps it even knows all that I’ve signed up to, and recognises overlaps/conflicts in the commitments…
And I think it should cost streaming fees, somewhere between $1 – $10 per month. At least that’s what my market survey of one indicates people are willing to pay….
Now where’s my link to kickstarter….