How the heck are you today?
Are you having a great week so far?
Today’s round up is my #AIChat session with Randy Lariar (his Twitter handle? @Lariar). Randy works at EY. He’s the manager of the Advanced Analytics and Data Strategy department. He also blogs about data issues here: https://lariar.blog/
Today’s topic?
Data strategy for financial services industry when it comes to using artificial intelligence.
If you think about it, Randy’s lessons can be applied to the media and entertainment world and other industries as well.
If I were you, I’d grab a pencil and paper and start taking notes.
Finally, mistakes are mine.
———-
**** It’s 2PM EST. Welcome to #AIChat.
My guests help you make sense of the #AI world in 10 to 15 minutes.
See the pinned tweet for the #AIChat rules.
Please welcome my guest @Lariar ***
— Nick Tang (@nickhtang) October 24, 2019
Q1 @Lariar: What type of AI data strategy does the financial services industry need to succeed? #AIChat
— Nick Tang (@nickhtang) October 24, 2019
A1 (1/5)
Strategy combines having a working knowledge of where you are today and plan for where you want to go. That plan should be informed by your organization’s unique strengths and the dynamics of the market you’re in. #AIChat— Randy Lariar (@Lariar) October 24, 2019
(2/5)
In Financial Services, just like most industries, it is clear that most strategies will involve AI and that AI needs good, clean data.#AIChat— Randy Lariar (@Lariar) October 24, 2019
Guess #AI could now make better decisions than brokers, also make changes and tweaks in milliseconds !!! #AIChat
— Jan Barbosa 🐝 (@JBarbosaPR) October 24, 2019
Faster <> better. All comes down to the data used for training.
Risk is a big topic in finance and it’s not going away with AI. Maybe increasing (e.g. flash crashes, cyber threats)
Key is putting AI to work in controlled ways
— Randy Lariar (@Lariar) October 24, 2019
(3/5)
Good – Data that is structured/standardized so that ML and other models can provide actionable insights
Clean – Data that can be trusted to reflect the actual facts of the business, market, customers, etc.#AIChat— Randy Lariar (@Lariar) October 24, 2019
(4/5)
There’s a saying about us management consultants – “we borrow your watch to tell you the time” – true we do a lot of listening because every situation is unique. We structure what we learn and provide an outsider’s perspective that generally adds a lot of value.#AIChat— Randy Lariar (@Lariar) October 24, 2019
(5/5)
The kind of data strategy needed starts with “it depends” – but it usually involves
✔ Strengthening data governance and quality
✔ Creating an organized store of good, clean AI data
✔ Enabling fail-fast in enterprise settings
✔ Most of all staying curious #AIChat— Randy Lariar (@Lariar) October 24, 2019
Can you explain the data governance side?
— Nick Tang (@nickhtang) October 24, 2019
Data governance, when done well, is the secret unlock to high-performing data in general and for AI specifically.
What’s a customer? Is it a person, an account number, an email address on an order form? If you want to build AI about customers you need to be clear on this!
— Randy Lariar (@Lariar) October 24, 2019
Same thing for every field that goes into the model. Where does it come from? What happens when there are conflicts from different systems. Who decides. Etc. Governance is a plan and operating model to work these out.
— Randy Lariar (@Lariar) October 24, 2019
Join me and my guest @SerrahL for #AIChat on Thursday, November 28 at 2PM ET! Topic? TBA!
Have a great week!
Try to be good!
— Nick Tang (@nickhtang) October 24, 2019