In this episode, Jason speaks to Stephen Galsworthy, Head of Data at TomTom about the gradual integration of artificial intelligence (AI) into data products to create a better user experience
Listen to this episode on Spotify, iTunes, and Stitcher. You can also catch up on the previous episodes of the Hub & Spoken podcast when you subscribe.
For more on data, take a look at the webinars and events that we have lined up for you.
Businesses already use AI to turn raw data into valuable insights on a regular basis. However, when you combine AI and data you create products that can enhance end-user experiences for the customer. In turn, happy customers will keep using these products, providing more data you can use to tweak and improve upon.
[00:35] Stephen’s background in mathematics and what sparked his initial interest in data
[03:54] How TomTom navigated the shift from hardware to software and AI
[06:51] Why data should be productized instead of being viewed as separate parts
[10:30] Combining AI into data products
[15:20] How to determine you are on the right path
[20:30] Where to align your data team depending on the maturity of the business
[27:30] Challenges associated with integrating AI with data
[34:32] The restrictions of AI
Even though there is a small subset of people who would enjoy having access to all the data to read and interpret, the majority of people just want a simple and effective user experience. For example there are many different data points for electrical appliances around a house. Instead of putting forward all the different data of kilowatt usage and other variables for each appliance, AI can simplify this process. By combining this with the ability to perform certain actions such as shutting off certain appliances or altering how they function, AI can be responsible for doing the heavy lifting. Now, with the power of AI all the user has to do is answer whether or not they would like to save energy with the push of a button.
The process to determine how data integrates with AI to create a user experience is not always seamless and involves a lot of research. The success of an AI system is primarily based on what the user requires, and what is possible with the infrastructure you have. Between these two factors a flywheel occurs: data is used to create a better user experience and therefore more users who enjoy the updated experience provide more data, creating an even better user experience.
While it is good to have a future plan in place, it is also unreasonable for a young business to replicate the systems and methodology of a more mature business. A young tech company simply cannot take a fully built infrastructure and transplant it into their own business expecting it to work the same. Each business needs to build their strategy and infrastructure from the grounds up, allowing customers to provide valuable feedback along the way.
The best way to improve the user experience is by gathering data directly related to what the user actually wants. Sometimes data experts get wrapped up trying to implement and streamline models that end up being completely irrelevant to the end experience. By starting from scratch and creating a foundation based on what your user wants will end up being more beneficial to the business in the long-run.
There are two main challenges that data teams face when trying to integrate AI into data products – mindset and complexity of engineering.
Mindset affects stakeholder and data teams as the integration of AI and data requires a more agile approach. At the end of the day it comes down to company culture and the acceptance of ‘failing fast’ as data teams try to find and implement better solutions.
In terms of the complexity of engineering that may be involved, knowing the limits of skills and capabilities is a good starting point as you begin to map out your strategy and build. From there, you can work out where upskilling or onboarding of new members may be required to help with AI capabilities and expand the data team’s impact.
The ability to combine AI and data together to create a product is more difficult logistically but it also carries a greater responsibility. When you develop a data product the raw data transforms from being an educational tool that facilitates decision making into something an end user interacts with. This transformation increases responsibility as there are implied guarantees that come with a data product, including the ability to function and perform as needed along with regular upgrades and maintenance.
When it comes to using artificial intelligence in business, there is no shortage of opinions on how and where it should be used. One of the most potent places that AI can be used is in combination with data to help create better user experiences that ultimately help to fulfil the goals of a business.