This CEO's Important Tip: "Nothing in AI Is Ever 100%"
AI + You
In a single sentence or statement, tell us what you do.
As Founder and CEO of Mobilewalla, I am responsible for setting the business and technical strategy of the company and for providing the leadership for us to achieve our revenue goals.
How do you define AI? (Or, what’s your favorite definition of AI?)
In its most basic sense, AI is using machines (i.e. automation) to perform decision-making tasks that were traditionally performed by humans.
From a technology perspective, AI is a set of capabilities, like algorithms and data, that when used together creates an automated decision-making system.
Sundar Pichai, Google’s CEO, has stated that, “AI is probably the most profound thing humanity has ever worked on.” Do you agree? Why, or why not?
I disagree with this statement because it takes too narrow a view of broader human endeavors—poetry, art, music, cinema—all that moves human beings are profound as well.
In the area of human technology interaction, then yes, I would agree that AI is very profound.
How did you get started in AI?
I am a computer scientist by training. I majored in computer science in graduate school and then worked in computer science departments at large universities. My PhD research was in the area of large-scale data management, which has been my focus.
I have always worked on, and enjoyed, solving complex problems with data. One such problem is AI. As it turns out, much of the complex data manipulation that I innovated is extremely useful for AI.
The net is that I have always worked in an area that was essential to driving AI.
"Nothing in AI is ever 100%, and in many examples the output of AI needs to be interpreted and acted on by humans."
What’s your favorite example of AI in your daily life that most consumers take for granted, or don't even realize is made possible by AI?
An implementation of AI in real life that I find to be very useful and effective is image recognition. Lots of things we do daily are based on image recognition. For example, fingerprint-based identification on your phone.
Image recognition is not exact. Things never match 100% (like your thumbprint). The matching is all predictive and AI-driven, and the algorithms determine if the match is close enough.
What excites you most about AI?
What I find most exciting about AI is the ability to scale mundane tasks that take humans a lot of time and effort.
What worries you most about AI? How could it go wrong?
Today, AI is being applied by organizations in many different areas. These application areas have different error-tolerances. AI is probabilistic and predictive and is better suited to situations that can tolerate more ambiguity (like marketing), not life threatening situations where there is no room for error (like medical decisions).
Nothing in AI is ever 100%, and in many examples the output of AI needs to be interpreted and acted on by humans.
In summary, my biggest worry about AI are the secondary effects of applying it to areas that have low tolerance for uncertainty.
What do you think is the biggest misconception about AI?
That it will remove humans from decision-making.
What skill or trait do you believe has the greatest chance to remain uniquely human for the foreseeable future?
Certainty. In the end, AI can only offer options with likelihood, choices with weights. AI outputs solutions from things you can model where you are making a choice.
A human can conceptualize abstract things, AI cannot.
What is a recent advance in AI that blew your mind?
"Marketers need to have clear goals for an AI implementation and a good use case that is a fit for this technology. They need to make sure they have enough of the right data necessary to build an effective ML model and they need to make sure they have the right people and technology resources."
If you were the dean of a business school, what is one thing you would do right now to start better preparing students for the intelligently automated future?
So many business processes incorporate AI. It is hard to go into a business and be a manager without understanding the processes you are managing. I would specifically create a course on applications of AI in business. Not a computer science course, but a case study-oriented course that brings in AI-driven examples from different industry segments and gives a business leader the knowledge and understanding they need to effectively use AI to solve their business problems.
Marketing + AI
What advice would you give to marketers looking to pilot AI in their organizations?
Marketers need to have clear goals for an AI implementation and a good use case that is a fit for this technology. They need to make sure they have enough of the right data necessary to build an effective ML model and they need to make sure they have the right people and technology resources.
What is the biggest challenge marketers should plan for as they scale AI?
Access to the breadth and depth of data necessary to power AI. AI is very data intensive and most organizations' internal data will not scale to meet the needs of AI. It is important for marketers to understand the internal data they have access to and to look externally at data sources that can enrich what they currently have and allow them to expand their AI efforts.
What’s one marketing job you see AI fully automating and eliminating in the next five years?
I don’t see AI fully automating and eliminating marketing jobs. There is just too much nuance in effective marketing to rely on everything being run by machines.
What’s one marketing job you see AI creating that doesn't exist today?I would see the evolution of a Marketing Scientist role: Someone who understands both marketing and the data science process. This role would be a bridge between the marketing and data science teams.
What can marketers do to ensure the ethical use of AI in their marketing?
Marketers need to ensure the use of data with permissible purpose and consent.
They should also be relying less on data sampling and try to get larger, anonymized, population-level data sets.
How can brands achieve personalization without invading privacy?
Move from viewing a consumer as an individual and instead as a part of well-organized groups (like fitness lovers, gamers, frequent diners, travelers, etc.).
AI can help with granular, highly accurate and nuanced, multi-dimensional segmentation which still allows a high degree of personalization, while simultaneously abstracting away consumer identity to preserve privacy.
"I don’t see AI fully automating and eliminating marketing jobs. There is just too much nuance in effective marketing to rely on everything being run by machines."
Voice assistant you use the most?
- Google Assistant
- Don’t use voice assistants
First publicly traded technology company to reach $2 trillion market cap?
Preferred cloud for building AI solutions?
- Amazon Web Services (AWS)
- Google Cloud
- Microsoft Azure
- Don’t use or prefer any of them
What does an AI agent win first (or at least share with a human)?
- Nobel Peace Prize
- Won’t win any of the above
Favorite AI movie?
Favorite AI book?
Favorite piece of AI content you've created that you'd like to share with our readers? Include any relevant links.
About Paul Roetzer
Paul Roetzer (@paulroetzer) is founder and CEO of PR 20/20, author of The Marketing Performance Blueprint and The Marketing Agency Blueprint, and creator of The Marketing Artificial Intelligence Institute and Marketing Score. Full bio.