We have known throughout human time that their is wisdom in crowds. Hunters followed paths created by their ancestors , ignoring the paths that didn’t help and focussing on those that did. We have ways of developing things from arrowheads to computer chips by learning from previous ways of doing things.
It’s the same in finance. To take an example, theory tells me what the average person has invested in for his or her retirement. I can construct a fictional fund and create a fictional unit-price track, I can compare the progress of my investments with this fictional track and see if I have done better or worse than the fictional average.
But if I repeated that process, not just with my data , but with the data of say 5 million others, I would find that the fictional price track might not be the average, in fact I could create a “non-fictional price track” which would be the average daily experience of all who are investing for retirement.
This is an example of artificial intelligence at work. We start with real intelligence and test it against reality and then end up with a bigger broader , more accurate construct, that is really fit for purpose. “Artificial” does not do this intelligence justice, this intelligence is based on the wisdom of the crowd,
The pathways we tread
I’ve been looking at a technology platform called Abaka. They’re the people who’ve trade- marked the phrase Artificial Financial Intelligence that you see at the top of this blog.
Which is comforting reading if you are looking to take ideas to market without the money to build your own platform.
What a technology platform does, is replace the human voice with the intelligence of a crowd of people who have come before you. Now you may think that crowd wise or you may think them foolish. “Herding” is a most dangerous behaviour (ask the Genezarine swine who collectively through themselves off a cliff into the sea.
“In and with” confidence
But normally we tread the trodden path because people like us do.
And the point about conversational AI is a good one, provided we trust the bot we’re sharing with, we talk with them in and with confidence
And we’re prepared to get pushed around by a bot,
So long as the bot’s talking our language
And so long as it’s a nudge not a barge
Incidentally – all these images are nicked off Abaka’s website
But back to pathways
As many readers will have experienced, the Lady Lucy plies her way throughout the spring, summer and autumn, up to Sonning. It passes Wargrave as it does where the river loops almost into a giant horseshoe.
For sure, were this 2520, that horseshoe would be no more and the river would go straight from Shiplake lock – down river to Marsh lock, but for now we enjoy the meander.
Sometimes it makes sense to go with the flow, sometimes, when we have a map to hand , we choose to go the direct route. Artificial intelligence can tell us what the quickest route will be, but human intelligence may over- ride – that loop is a wonderful boating experience!
That is why a conversation with a bot is instructive but not definitive. Many of us will choose the less effecient but more enjoyable way and many will choose to be guided by the hand of a human.
Hope for me yet!
This interaction between the effecient and the suitable pathway is what AgeWage will be testing over the coming months.
I’m impressed that Abaka have recently raised $6.2m from the market to deliver the platform to people like me. And I’m really pleased that they are thinking like me about how we can bring AI to the rescue of people who need to turn pension pots into retirement plans.
Ok – so this is not the language of the common man, but this is stuff that ordinary people need to make the difficult choices and avoid the perils of the Strait of Hormuz.
And there’s enough in this to leave a revenue stream that makes the enterprise viable.
Because if all this kit doesn’t amount to a definitive course of action , imprinted in the user’s financial DNA, then it won’t have worked.
An API (officially an “application program interface” is the kit that brings data together.
In the vast flow of the Lower Thames is the water from all the tributaries from the upper and middle Thames which have been integrated by the Daddy river .
When you use an App that works, it is like the Thames, it brings all the little streams of information and of knowledge to it so that you can eventually get to Reading, or Windsor or London or even the English Channel.
I know all the little rivers. There’s the Ferret and his system to work out what we can get from universal credit. There’s Retirement Line with their expertise in getting us the best annuity rate and there’s a local expert like Pension Bee that can bring your pensions together with the help of the bee-keepers. There are many undiscovered streams and culverts which can integrate to a central flow , and data and water have a lot in common!
Which I guess means that small but bright organisations like AgeWage , can plug into bigger and more developed organisations like Abaka and deliver the kind of dashboards on the kind of apps , that get people excited about their pensions.
We know what front ends look like, they’re what make people buy and they have to be as glamorous and sexy as Harrods’ shop window.
Artificial intelligence is about turning the eye-grab into getting things done. AI only adds value if it “imprints into the user’s DNA” a desire to take action so strong that something gets done.
Of course that “something” could be good or bad. But if we trust our intelligence to deliver pathways that are based on well-trodden ways, we can set AI to work for us as Google Maps works for us, as Siri works for us.
I sat with my doctor in the week before Christmas and we discussed aspects of my health: four times my doctor asked his phone the questions he could not answer himself. Each time he chose the credible answer presented by Siri and each time I asked myself why I needed the doctor.
That doctor gets paid a lot of money, I don’t begrudge him a penny!
AI has sadly become a marketing term, divorced from the reality. I worked in AI in the past, including the Governments Alvey project in the 80s, and used to program in AI languages like Prolog and Lisp. Most of the people who have worked seriously in AI over past decades, point and laugh at many of things labelled as AI today. There is some incredibly impressive stuff being done using machine learning and big data in many fields and the IBM Watson spin-offs are fantastic. But most of the stuff that’s being pushed is just a combination of simple rule-based stuff and document assembly. I think the stuff we do at Ferret is clever but it’s not AI. In finance and law there are fundamental difficulties in the way of applying genuine AI, not least because the way that rules are drafted tends to include defeats and redefinition of terms which can cause rule engines to implode.
In brief, before acepting a ‘magic box’ get someone who knows about the area to ask some detailed questions.