Historical data is valuable for training AI models. User-held historical data could change the game by empowering consumers with truly personalized AI.
Many parties are trying to gather lots of data so they can utilize it. Today we see all kinds of services that are offered for free to acquire consumer and business data – as the saying goes, if you don’t pay for a product, you are the product. But there is another interesting dimension to this – gathering historical data, which can be also very valuable for things like training analytics and AI models.
Some years ago, I talked with a friend who worked for Google. He asked if I knew how to convince Google to acquire a company for a good price. The answer of course depends on what that company has that Google finds valuable, or that it doesn’t already have. Put simply, Google doesn’t buy companies to acquire their excellent data scientists, their algorithms or their service – Google has plenty of these and can get more easily.
What Google does want is a huge cache of historical data it doesn’t already have, and it is ready to buy companies that have interesting historical data.
Data trading is a big market, but quite often it is a gray or black area and not very transparent in terms of how the data has been collected and whether privacy is respected. As an Uber data leader once commented, it is actually very hard to buy data. There are companies that want to sell data, but you don’t want to do business with those companies. And there are companies whose data you would like to buy, but they don’t sell it.
User-held historical data
Data is like fuel for analytical models and AI. You cannot build and develop models without data. The more data you have, the better models you can make. So, you often need more and richer historical data. So, how do you get hold of it?
User-held data models offer an interesting new opportunity for this. In this model, it’s not about companies buying individuals’ data, but rather individuals utilizing their own data and getting better models and tools for their own use.
In practice, this means people can link to or download their own data from different services, from social media and wearable devices to healthcare and finance. Then they can upload this data to their user-held data accounts, where they can be combined. Thus, you can basically have all your wearable, exercise, health, social media and finance data (and any other personal data you have) in one place that can be used for your personal applications and services. So, to take an easy example, your wearable device can make use of all kinds of data, not just the data specific to that app.
Many services already offer APIs or other means to access your data, and in many countries, regulations (e.g. GDPR in Europe or CCPA in California) requires service providers to offer this option to their users. Some services make it easy, some more difficult. But user-held data companies also offer better tools to make all this easy.
A unique opportunity
Compiling and combining different historical data sets presents a really unique opportunity to gather lots of historical data – not only for one service, but several services – and build better analytics and AI with it.
What makes this more unique is that these tools and AI models are not for companies to improve their business, sell more products and services or profile customers more accurately. These tools are for enabling individual people to have their own tools and personal AI. The data is not given to other companies or service providers – individuals can have services that run in their own environment (i.e. a personal cloud or device).
In practice, this could enable all kinds of personal AI services for our daily activities, health, lifestyle, finance and purchases. These tools can learn from our behavior and activities and combine that data with external data to help us much better and more personally.
At the same time, individuals will have access not only to their service data, but also to the data they generate from the sensors in their clothes and from IoT devices in their home. This will be a powerful additive to their historical data cache. And when GPT tech offers better and easier user interfaces to really utilize this data, it will really enable more people to use these tools.
Empowering consumers with personal AI
Many companies that make AI and analytics tools (as well as their investors) have focused for years on making such tools for enterprises to collect and analyze data. But this latest development is changing the market totally.
It shifts the focus much more to offering consumers better tools and applications. It will change the privacy discussions from pure restrictions to empowering consumers as well as protecting them.
Together with the rapid development of AI right now, the opportunity to use personal data, including historical data, is a powerful disruptor for the AI market. It empowers individual people in a really new way to have useful and beneficial services with their own data. It will really enable people to work and live with their personal AI.