Rational AI: Demystifying Machine Intelligence in the Enterprise
AI is a pursuit, not a thing — Adam Drake
I attended the Rational AI event at MaRS in Toronto on April 5, 2018. I wanted to write about my experience and what I learned from the event.
My previous AI Interview Questions articles for Google, Amazon, Facebook and Uber have been very helpful to the readers. As a followup, next couple of articles were on how to prepare for these interviews split into two parts, Part 1and Part 2. Do visit them and provide feedback as well.
Acing AI aims to help people get into AI. My goal with these event experiences is to share learnings and help people on the fringe understand the value around these events. If I get good feedback I may continue writing more of these. Do let me know.
The event had three main parts, two talks presented by some amazing folks from the domain followed by an interview panel with execs from the startup domain in the field of AI. All this was followed by networking which was the best part.
No Silver Bullet: A Rational Framework to Evaluate the Risks and Rewards of Enterprise AI
The event started with Kathryn Hume who is the VP Product & Strategy at Integrate.ai presenting on this topic. She explained the use of AI in the Enterprise elaborating on how ML techniques have progressed over the years to fit into an enterprise setting in various verticals. She described two case studies to distill a Machine Learning Product Lifecycle from Design to Product Evaluation and highlighted the risks and learnings from the case studies.
There is often a difference between the quantity you want to measure and the one you can measure-Yonatan Zunger
Her case studies were based on the hands-on ground reality of working on Enterprise AI solutions and her talk was titled rightly so — “No Silver Bullet”.
I understood the depth and breath of Kathryn’s knowledge about AI when I networked with her after the event. She had some amazing insights.
Important links:
- Kathryn’s AI Podcast: In Context
- Kathryn’s Blog: Quamproxime
Developing Your Own “AI BS Detector”
Adam Drake was the second speaker who gave this fascinating talk. The ethos of his talk was around the fluff that Enterprise AI is today.
What startups were to the VCs in 2015’s, AI is to the Enterprise today.
Adam demonstrated the various questions one can ask to distinguish between AI and non-AI products in the Enterprise. One of the questions that stood out for me was, “Why is what you’re doing considered AI?”. Solving the same problem using a different tool does not necessarily make it ‘AI’. It was a good checklist to understand how AI has become a marketing heavy term today.
His parting thoughts were: “When you hear AI, be suspicious”.
Important Links:
- Personal Website: https://aadrake.com/
Surfacing Deep Customer Value with AI
This was a panel talk with some amazing entrepreneurial folks. Starting from the left, Steve O’Neil followed by the great panel as described below:
- Karan Walia (Founder & CEO, Cluep)
- Bryan Smith (Founder & CEO, Think Data Works)
- Steve Woods (Co-Founder & CTO, Nudge.ai)
- Stephany Lapierre (CEO, Tealbook)
Steve O’Neil from MaRS was the moderator for this talk. He asked some interesting questions about AI and data practices within these companies. The people on the panel had some really interesting ideas from their current startup playbook.
The varying backgrounds of the panelists and their companies were enough to demonstrate how different problems can be approached with the help of AI from so many different angles. For some companies data was helping build their product while for some data was the product. This has different challenges of their own.
Steve O’Neil mentioned that such events can become a regular occurrence at MaRS which was amazing to hear.
Networking after the event was the best part of the event where I got the most value. I met some amazing folks who were doing an interesting body of work in the field of AI.There were people who helped advice startups to people who build Wi-Fi for giant stadiums. I also got to ask some questions to the accomplished people who presented at the event. It was a great few hours of learning about how a real AI can help the enterprise.
Important Learnings(In a Nutshell):
- Enterprise AI have a very different set of problems than Consumer based AI(AI at Netflix/Amazon). Risks involved between different Enterprise AI problems vary considerably. Solving a business problem using AI without risk evaluation may prove lethal.
- Problems which require a 100% accuracy are not good candidates to deploy AI in the Enterprise.
- Data Scarcity limits the type of models which can be deployed in an Enterprise AI setting.
- Data Visualization and presentation, context setting for stakeholders and evaluating model fairness are paramount to making a project successful.
- It is important to distinguish between AI and non-AI to identify and solve actual AI specific problems in the Enterprise.
“Always remember in the Enterprise, the AI’s goal is to solve business problems.”
If you find this article useful, please share your thoughts in the comments below. Please clap on the article to signal me how much you like this article and if you would like me to write more of such articles.
