Skip to main content

Data & Intelligence

AI Summit: There is Nothing Artificial About Artificial Intelligence

Perficient is at the AI Summit in New York City and there’s a lot of great content and insights. Jeff McMillan, Chief Analytics and Data Officer at Morgan Stanley Wealth Management gave a great session about AI and what he’s learned. In many ways, his insight mirror what Christine Livingston has to say about AI in general.  There is no silver bullet. It takes a lot of effort and focus but if you choose correctly, you can gain a lot of value.

Jeff started by talking about his recent attendance at a San Jose Conference last year and every company had transformed itself from a mobile, data management, or other company into an AI company.  The problem is that half of what they pitch was nonsense. You don’t transform with AI in the space of a year.

Quote: this stuff is hard and complicated. Some of it works and some of it doesn’t

Current State of AI

It’s mixed

  • Too much of the conversation is about AI and not about what we are trying to achieve
  • Your biggest challenge is to separate what’s real from what’s not
  • AI has promise, there are no silver bullets. It takes leadership, lots of people, and effort to make this work.

The AI universe isn’t a mythical blob but rather:

  1. Old stuff rebranded as AI
  2. AI that actually doesn’t do what it says it does
  3. AI that works but doesn’t solve a problem you have
  4. AI that adds value but it’s way too complicated to use
  5. AI that works but costs too much
  6. AI that can add value but lacks required data (like buying a Ferrari but not having any gas)
  7. Real, usable, achievable, and affordable AI

Focus on the old stuff and the real usable opportunities.

Quote: Figure out what makes sense for your own business and move forward

Quote: just because I call it AI doesn’t make it AI. It may not be bad. It’s just not AI

Data Intelligence - The Future of Big Data
The Future of Big Data

With some guidance, you can craft a data platform that is right for your organization’s needs and gets the most return from your data capital.

Get the Guide

Our focus needs to be on figuring out what works, what doesn’t and how to move forward

Rebranded AI

  1. You need to be training your employees in basic statistical terms and analysis then you are failing.
  2. You still get enormous value from traditional data science
  3. Conduct structured test and control experiments to measure
  4. Create a culture of fact-based decision making and collaboration

Quote: just do this stuff and you will make more money

AI that Doesn’t Work

Keep in mind that AI does not think.  If anyone says that, they are either lying to you or they have no idea what they are talking about.   You want to ask vendors to show tangible results. You don’t have to be someone else’s R&D department

Quote: Just because something can be done, doesn’t mean it should be done

Experiments should be built around definable outcomes.  You need people to think this through. When it’s too complicated, you need a culture and people who can challenge and question.

Quote: always bias towards the simple.  By going back to basics you drive more business value

When It Costs Too Much

Try to consider all the factors required to successfully deliver your solution.  Engage with the business.  Ask yourself if there’s a simpler way to do it. You will save time and money by thinking first.

When you Lack the Required Data

It needs to be accessible, curated, and predictive in nature.  Good algorithms here will still give you bad results. This means the lack of proper data governance structure drives significant mistakes.

Advice: Get your data governance in order if you want to use AI and ML

What Brings Value in Financial Service?

  1. Anomaly detection.
  2. Robotics Process Automation. Relatively cheap. Can sit on legacy infrastructure
  3. Intelligent Assistants: this is great if it works
    1. You can create a really bad bot in 10 minutes
    2. It’s really hard to do natural language processing
    3. Technology is the least part of the problem. You need access to the experts
  4. Predictive Analytics. It’s a decent impact without huge cost
    1. You have to have your data set
    2. But they do help you see things differently
  5. Decision engines: High complexity but also very high impact

An Example of what’s working

In 2009 Morgan Stanley started the “Next Best Action” project It went through starts, stops, failures, and restarts. It’s goal was about democratizing access to algorithms.  It identifies the most impactful and relevant ideas for clients. It ranks them based on propensity, outcome, etc.

What it does: take the expertise of their experts and put it into the system. The algorithms prioritize it. Present to humans and let’s them apply their experience, intuition, and empathy.

Then it iterates again and again.  It gets better each day. It helps financial advisors bring personalized information and recommendations.

What made it work?

  1. Skilled data scientists
  2. Integrated technical infrastructure
  3. A collaborative cross functional team
  4. Historically accurate and accessible data
  5. Differentiated content
  6. Tight coupling of the algorithm output with your daily decision-making process. (human in the loop)

Keys to success

  1. There is no AI without real intelligence
  2. Start with the problem and not the solution
  3. Near term, the greatest value will come from augmenting decision making
  4. Implementing AI is hard. Set your goals modestly
  5. The winners in this space are not the ones with the best technology. There are a lot of other elements
  6. Leadership that is paying attention to AI and supporting it will help you drive success.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Michael Porter

Mike Porter leads the Strategic Advisors team for Perficient. He has more than 21 years of experience helping organizations with technology and digital transformation, specifically around solving business problems related to CRM and data.

More from this Author

Follow Us