“Often with data science ideas, it’s about how they’re presented. So the machine learning algorithm is only part of the larger solution. Which is, in itself, running as part of an IT architecture… And you need to understand that data scientists can’t be expected to understand all of that architecture.”
We had an engaging sit-down with the Head of AI and Machine Learning in Customer Solution Engineering at Dow Jones, James Bell. He shares personal and valuable insights about the importance of building communities for data scientists, taking a broader approach to change and the secret to finding sweet-spots for AI solutions.
Building Internal Communities for Deeper Insights
“It was creating a community internally that really helps make a difference. One of the things we set up was a regular group that meets… That helps data scientists talk to data scientists”.
“Someone presents their ideas, what they’re working on, and they get that broad aspect from the rest of the company that says, “Well, have you thought of this?” and that’s really helped people be successful. Because a lot of the time, dedicated data scientists are not very knowledgeable about business, and consequently, that has been a real help to them.”
A Broader Approach to Change
“Within each business unit there are AI people and machine learning engineers and data science researchers who are doing fantastic work. And so that needed to be respected.”
“I always think it starts at the top with values and processes and then setting the sort of guide wheels so that people can express their ideas in that business area without causing damage or danger. Because a lot of ideas are really good, but when you actually scale them up or you start thinking about how they’re going to impact the public, then you need to start thinking about the ethical and policy-based implications.”
“… If you have something that says you must look at data privacy as part of your AI design, what does that practically mean?… Then it’s about setting up a group that can talk and assist them in making their ideas come to fruition”.
Finding Manual Burdens to Build AI Solutions
“When I’m designing something, or when I’m coming up with a new idea for something, I always look at “what are we doing manually now?”… I have this saying that artificial intelligence occurs where there has been an overabundance of burdens. A manual burden. You should look for those manual burdens”.
“So for example, at KPMG, we were designing things to replace manual risk and compliance work, which is done by thousands and thousands of people in call centers… on behalf of big, tier-one banks. How can we move those into automations, while at the same time respecting the impact of automating manual processes?”.
Find Out More About Communities of Scale in Data Science, the Hidden Value of Manual Burdens and Plenty More!
Get in the mood for some interesting banter on this week’s episode.
And make sure not only to give it a good listen, but feel free to share any of your thoughts, opinions, comments or ideas with us!