My new post on LinkedIn gives a few tips on governance. How your special perspective can help you to build a well-organised ecosystem.
My new post on LinkedIn is about how looking for the right data is like looking for a needle in a haystack. But we found a way to match data generators with data users – even for unpredictable data needs.
My new blog on LinkedIn is about the organisations which will deal with all the data that our AIs will need to work well. How do we make sure these organisations do not stray?
One type of AI software uses neural nets to recognise patterns in data – and it’s increasingly being used by tech firms like Google and IBM. This type of AI is good at spotting patterns but there is no way to explain why it does so. Which is a bit of a problem when the decisions need to be fully accountable and explainable.
I could have called this post ‘One thing you cannot do with AI at the moment. There are many things that AIs are helping businesses with right now. But if your firm is going to use them then it’s important to know their limitations.
I remember doing my maths homework once and getting low marks even though I got the right answers. The reason I lost marks was because I didn’t show my working out.
Sometimes the way that the answer is produced needs to be clear as well.
It is like that with some AI technologies right now. There are types machine learning AI that are amazing at recognising patterns but there is no way to explain how they do it.
This lack of explainability can be a real barrier. For example, would you trust a military AI robot armed with machine guns and other weapons if you weren’t sure why it would use them?
Or in medicine, where certain treatments carry their own risks or other costs. Medics need to understand why an AI diagnosis had been made.
Or in law, where early versions of the EU’s General Data Protection Regulation (GDPR) introduce a “right to explanation” for decisions based on people’s data.
The problem is that for some types of machine learning, called “Deep Learning”, it is inherently difficult to understand how the software makes a decision.
Deep Learning technology uses software that mimics layers and layers of artificial neurons – neural networks. The different levels of layers are taught to recognise different levels of abstraction in images, sounds or whatever dataset they are trained with.
Lower level layers recognise simpler things and higher level layers recognise more complicated higher level structures. A bit like lower level staff working on the details and higher level managers dealing with the bigger picture.
Developers train the software by showing it examples of what they want to it recognise, they call this ‘training data’. The layers of neural networks link up in different ways until the inputs and the outputs in the training data line up. That’s what is mean by ‘learning’.
But neural network AIs are like ‘black boxes’. Yes, it is possible to find out exactly how the neurons are connected up to guess an output from a given input. But a map of these connections does not explain why these specific inputs create these specific outputs.
But on its own a neural network AI cannot justify the pattern it finds. Knowing how the neurons are connected up does not tell us why we should use the pattern. These types of AIs just imitate their training data, they do not explain.
The problem is that lack of accountability and explainability. Some services need proof, provenance or a paper trail. For example, difficult legal rulings or risky medical decisions need some sort of justification before action is taken.
Sometimes transparency is required when making decisions. Or maybe we just need to generate a range of different options.
However, there are some possible solutions. Perhaps a neural network AI cannot tell us how it decides something. But we can give it some operating rules. These could be like the metal cages that shielded production workers from the uncertain movements of early industrial robots. As long as a person did not move into the volume that the robot could move through then they would be safe.
Like safe paces to cross a road. Operating rules would be like rules of warfare, ground rules, policy and safety guidelines. Structures that limited the extent of decisions when the details of why the decisions are made are not known.
A similar idea is to test the AI to understand the structure of what sort of decisions it might make. Sort of the reverse of the first idea. You could use one AI to test another but feeding it huge numbers of problems to get a feel for the responses that it would provide.
The antelope image that was generated by the AI shows a little about how the AI software considers to be separate objects in the original picture.
For example, the AI recognises that both antelopes are separate from their background – although the horns on the right hand antelope seem to extend and merge into the background.
Also, there is a small vertical line between the legs of the left hand antelope. This seems to be an artefact of the AI software rather than a part of the original photo. And knowing biases like that helps us to understand what an AI might do even if we do not know why.
But whatever the eventual solution, the fact that some AIs lose marks for not showing their working out highlights that there are many different types of AIs and they each have their strengths and weaknesses.
My new article on Linkedin Pulse suggests how firms that make IoT products and services could build their own IoT ecosystems and how they could persuade other firms to join.
My new article on Linkedin Pulse looks at the implications of how the sensors on modern cars can tell manufactures where you drive, how you drive and when it might be about to breakdown. This data could be a massive help to car repair garages, insurers and other firms. But manufactures might want to charge for it and this might increase the cost of servicing your car. So how will firms partner together?
My research suggests strategies for building Internet of Things (IoT) ecosystems in the car industry and other IoT product industries.
The success of the Internet of Things (IoT) depends on sharing data between different devices but it needs to be done appropriately. But there are some brand new technologies that might support appropriate data sharing and they can do it on a massive scale.
Sharing (data) is good for us. The more we share then the more perspectives we have on any given consumer or any given problem. In retail this means getting a better understanding of what customers need even before they do. For the IoT this means combining the sensors and abilities of many devices to get a fuller picture of what’s needed and having more options on how to do it.
Firms that manufacture IoT products only get direct data from the customers that use their products. One device only ‘experiences’ a part of a customer’s life. To get a fuller understanding of a single customer’s life different devices need to share data, preferably devices that help the same customers in different ways.
The classic example is a shopping app on your phone that uses data from your refrigerator to suggest recipes based on what you actually have in stock. Or a firm that uses data from all the devices that it has sold in the past to give advice to specific customers, like Waze uses crowd sourced data on traffic jams to warn its users.
But really useful data can come from external sources.
Just like good feedback and useful suggestions, external information is more likely to give you a new perspective. So devices that share data with many other devices will tend to get more useful information and make better decisions. Decisions that are either more personalised, because of complementary information about the same user, or decisions that are chosen from more options, because of information on more user situations.
However, the problem is that firms don’t like to share with other firms for commercial reasons. More importantly, firms can’t just share customer data with any firm that they want to because of data protection legislation. How would you feel if your data was shared with anyone and everyone?
The problem is that sharing data makes for better user services but sharing data must be done appropriately.
Here’s the solution: if you want to share your data, meaning sell it or even just release the value in it for free, then first you need to stop talking about ‘data’. Be more specific. Being very specific is the key to reducing data sharing sensitivities.
If you ask me whether you can share all my data with everyone, everywhere and let them use it as they choose and pass it on to whoever they want to, forever. Then I’m going to get a bit tense.
But if you tell me exactly how you want to use it, why you need to use it, who else might get a look at it and what they are entitled to do with – either the data or the products of the data – and how long this is for. Then I will be much more open to your suggestions.
I still won’t give you my medical data so that your algorithm can charge me a higher insurance premium. And I hate it when it feels like my data is being used to sense when I will pay more for dodgy science fiction books on my Kindle or more expensive airline tickets.
But generally speaking a highly specific definition use of will remove a lot of my reservations. And real anonymisation will remove all of them. Q: who owns this anonymous data? A: dunno.
So how can we apply this to the IoT one devices at a time?
It is relatively easy to run a loyalty card programme because it is centralised and largely automated. But the IoT needs to be decentralised. If every device asks you or its manufacturer for permissions and instructions every time it needs data – or when is asked to supply data – then the IoT will not get very sophisticated.
So how do we ensure highly specific data sharing? Well, we need to design smartness into the IoT. So let’s use smart contracts.
Smart contracts are a mixture of contract law and software. The rules that govern a relationship between two parties are embedded in the software that facilitates the same relationship. To put it another way, the contractual terms are embedded in the thing that they govern. So they control it to help ensure that agreements are kept.
Software is made of rules and instructions for use but contracts also need proofs of things like identity, ownership, endorsements, permissions and rights to do things. And they need to be cryptographically secured, decentralised and machine readable so that every device can automatically ‘negotiate’ with other devices as required.
Fortunately all this can be written into blockchains, the distributed and cryptographically encoded databases that underlie digital currencies. In principle, the data sharing rules for each device can be broadly pre-specified; for consuming third-party data as well for supplying data.
Smart contracts could automate a multitude of interactions between IoT devices and pre-specify what constitutes appropriate behaviour in different circumstances. Automation is important because the more devices the better and the more granular the interaction then the smoother the service. Pre-specification is important because you do not know which devices your devices will meet.