The Power of Starting

Bertha Benz, Story of the First Cross Country Car Trip, and Your Journey to Artificial Intelligence (AI)

From Mercedes Benz

Lately, I have been traveling to Europe, specifically Germany, quite a bit. After all, we are expanding our business there, and travel comes as collateral. On the bright side, we get to see new places, meet new people and learn from them, get familiar with new cultures, and hear some inspiring stories.

Recently, as I was traveling to Wiesloch, Germany, I came across one story that stayed with me: Bertha Benz and her 66-mile car trip in 1888 from Mannheim to Pforzheim, Germany. Since we celebrated International Women’s Day this week, I thought of writing the story of the superwoman.

Interestingly enough, we can learn a lot from her as we start our journey to implementing Artificial Intelligence (AI) in an organization.

An Unheard Love Story

“Who you marry is the single most important career decision you make.” — Sheryl Sandberg, Lean In

Sitting at my Washington, D.C. apartment, as I look up and see thousands of cars crossing the Potomac River or taking the GW Memorial Parkway to the Reagan National Airport, it is quite a daunting task for me to imagine Patent-Motorwagens Model III, the world’s first production automobile, sitting lonely and useless in the humble home of Bertha and Carl Benz in Mannheim, Germany in 1886. It is even more challenging for me to fully appreciate the heroic effort of Bertha Benz.

Bertha Benz

Young Bertha Ringer met Carl Benz in 1869 when she was 20 years old. According to the Mercedes-Benz classics,

“When Carl brought the subject around to the horseless carriage on which he was working, young Bertha lost her heart to him.”

Young penniless engineer, Carl, was not a natural match for Bertha, who was born to a wealthy family in Pforzheim in 1849. But Bertha was determined to marry the visionary, rare breed of a designer-engineer. I think she was dying to know if we would ever drive a horseless carriage, and knew that Carl would get her there. She fought her fight, married Carl, and invested her entire dowry into the development process. Carl didn’t break Bertha’s heart. He came up with hundreds of small and not-so-small inventions (such as gasoline-powered two-stroke piston engine), and several prototypes ultimately converged into the Model III.

Carl Benz

A patent registered on 29th January 1886 formally made Carl Benz the inventor of the automobile, at which point the story took an interesting turn. Bertha and Carl’s dream was not so obvious to everyone else. No one was interested or confident about the horseless carriage and its ability to travel long-distance.

After spending two miserable years with the thought that horseless carriages would never happen, Bertha decided to do the unthinkable: travel 66 miles in the Model III to Pforzheim to meet her mother.

“In the early hours of a fine August day in 1888, together with her sons Richard and Eugen, and without the knowledge of her husband, Bertha Benz took to the road with the Benz Patent Motor Car. She was undeterred by the fact that some stretches of the roads, which were normally used only by horses and carriages, were anything but suitable for the automobile. Lack of fuel, clogged valves or wiring chafed-through to breaking point — she found a solution to every difficulty on the journey. She resorted to a garter, a hat-pin, and plundered the ligroin stocks of pharmacies along the route. Even when the fuel ran out completely outside Wiesloch, and the Motor Car had to be pushed for several kilometres, she was not too proud to get down herself and help.”Mercedes Benz Classic

For many things in our life, all that we need to do is hit the start button. Half of the battle to realize our dream is that vision, courage, and willingness to give it a start. The other half are some principles that Bertha followed, some behaviors she exhibited.

In my current role, I see a tremendous willingness among Executives all around the world to make their organizations AI-driven, but I also see a hesitation as they don’t know how or where to start.

Let’s see what we can learn from Bertha (and Carl Benz) as we start our journey to the AI-driven organization.

Your Journey to Artificial Intelligence (AI)-Driven Organization and Learning from Bertha Benz

“Mood is noxious. Noise is costly to the Organizations, which are essentially factories for making decisions.”— Daniel Kahneman, Nobel Laureate Behavioral Economist

Before I get into how we start an AI-driven organization, let me first explain what we mean by AI-driven organization.

What Kahneman described as “noise,” is a serious concern for any organization as they make thousands of business decisions every day. Different experts making different judgments under the same circumstances with the same data — this random variability is “noise”. Kahneman gives the example of a radiologist examining images of the signs of cancer and how the mood of the radiologist may change the judgment. Human beings are heavily influenced by emotions, moods, and biases as they make decisions, which Daniel Kahneman documented throughout his book “Thinking, Fast and Slow.” Other behavioral economists Richard Thaler, a pioneer in the field, winner of 2017 Nobel Prize in Economics, and Dan Ariely have years of well-documented research and literature on this topic.

Machine learning algorithms and AI have two advantages over humans: first, their ability to process large, complex, unstructured data; and second, their emotionless way of making predictions and not falling for biases. And they make the whole process incredibly fast.

An AI-driven organization fully utilizes machine learning algorithms to process data and generate predictions, ultimately using those predictions as critical building blocks for evidence-based decision-making in every corner of the organization.

Some simple examples are the automatic or nearly-automatic issuance of a credit card, an auto insurance policy, or processing of renters’ insurance claims. At DataRobot, we use automated machine learning to solve thousands of such prediction problems across a variety of industries, including critical credit scoring predictions needed in micro-finance organizations, plant growth prediction in farming, and insurance claims management.

Now, let’s see how we get there following Bertha’s path to success.

Learn the Basics and Co-own the Vision: When Bertha heard about Carl’s vision of horseless carriages, she fully absorbed it, learned the basics, and started co-owning it. It is evident from several problem-solving efforts throughout her journey that she understood the working principle of a car. It is enormously important for Executives to understand the basics of AI and machine learning and genuinely co-own the vision for AI with their technical teams.

In 2018, anyone with an internet connection can learn serious cooking from Gordon Ramsay or Wolfgang Puck or writing from Malcolm Gladwell (in case you don’t know what I am referring to: Masterclass). Executives should dedicate a couple of hours every week to learn the building blocks, business applications, and economics of AI and machine learning. In some organizations, Executives are keeping their bright, and humble recent STEM hires as their “tech mentors.”

A Stake On the Table: I haven’t seen a successful AI initiative without executive sponsorship. Success or failure of your AI initiative should leave a mark on your year-end performance card. Bertha invested all her dowry in making their vision a reality. Just like Bertha, as an Executive, you need to give financial and moral support, stay involved, and let your people know that you have tied your success and failure to the AI initiative. This commitment changes everything.

Fast Iterations: One of my all-time favorite Harvard Business Review articles is “Special Forces Innovation: How DARPA attacks problems.” According to the article, DARPA has an “unwavering commitment” to work on projects motivated by well-defined, use-inspired need. This article also highlights “fast iterations” as one of the building blocks of DARPA’s project management. As the private sector aspires to become AI-driven, they should follow DARPA’s preference of use-inspired projects and nurture an environment of fast iterations.

“[Successful execution of these projects] involve fast iterations. Planning should be light and nimble. Progress can be assessed by tracking iterations to see if they are converging on goals, revealing dead ends, uncovering new applications, or identifying the need for unforeseen scientific advances.”- Special Forces Innovation: How DARPA attacks problems, Harvard Business Review, 2013

Carl, as a genius engineer, followed this fast iterations and prototyping approach. He built several versions before Model III. On the other side, Bertha had a very light plan for her entire trip. But she was mentally and technically ready to solve problems along the journey.

Importance of Short Cycle: Developing a 2-year plan for AI initiative is equivalent to a harakiri for any organization. The world is changing very fast, and no one is going to wait for two years to see the outcome of your initiatives. Build a 3-month to 6-month plan. The pressure of time will force people to deliver results, and it doesn’t matter if everything is not perfect.

Modern machine learning platforms are incredibly fast and can talk to any system using a technology called API (Application Programming Interface). A well-designed platform is a great story-teller that makes it easier to communicate complex mathematics across a wide variety of audience. The recent progress in technology and user experience allows organizations to deliver results in bite-size and in short-cycle.

The very fact that Bertha made the trip with the car that was sitting there for two years made all the differences. It was not a beautiful journey for her, but it was enough to change the course of humanity.

Creativity and Humility: To start a new city, Uber GMs hop in the yellow cabs all day long and keep convincing them to join Uber. Uber GMs are graduates from top business schools and often come from top management consulting or Wall Street firms. But they know the value of doing little things. As I spent my last three years at DataRobot, one of fastest growing AI companies, and developed the insurance practice to a multi-million dollar business in matter of 2 years, I diligently followed this principle. Some of our largest deals started with most humble beginning e.g. a 5 min pitch. If you are doing something for the first time, very likely you will have to look beyond your existing solution space and as an Executive, encourage your team to do so. Creativity and humility make the solution space infinitely bigger.

“Bertha resorted to a garter, a hat-pin, and plundered the ligroin stocks of pharmacies along the route. Even when the fuel ran out completely outside Wiesloch, and the Motor Car had to be pushed for several kilometres, she was not too proud to get down herself and help.” -Mercedes Benz Classic

[tech alert] At AIG, back in 2013/ 2014, we wanted to build an automated ranking system of all incoming commercial insurance policy applications (a problem commonly known as “submission prioritization”). AI will help the underwriters (UW) sorting their stack of policy applications every morning in decreasing order of “value,” and they will start reviewing from the top of the pile. This would maximize the ROI of UWs time and effort, and AIG would have better chance to write the most valuable policies in the market (big ticket problem as AIG gets thousands of applications every day and has hundreds of UWs across many underwriting divisions). We built the AI-system using R (a commonly used open source language that supports many advanced machine learning algorithms) but got stuck in the implementation as the front-end only can take SQL (Structured Query Language) through a legacy software. We did a creative hack at this point: wrote a code that would automatically translate all R code to SQL code (a process known as metaprogramming). We had hundreds of models covering all UW divisions — so this automation made the integration super easy. With DataRobot like software, these large-scale development (model factory) and multi-platform integrations are significantly effortless these days, but there will always be a newer set of problems that we need to solve with creative hacks. [end of tech alert]

Paint a Story: Bertha or Carl didn’t have a PR. But Bertha’s trip somehow connected with common people that influenced a lot of changes. She didn’t take 1000 rounds in a large open field but went to her mother’s home and came back. She connected with the people by showing the utility of an automobile.

“while some of the onlookers would prostrate themselves on the road in prayer, fearing this “smoking monster” as a harbinger of the Last Judgement, others asked for a test ride.”-Mercedes Benz Classic

As you start your journey to AI-driven organization, early success and excitements are super critical. So you have to choose problems carefully.

Don’t select a problem that Google is trying to solve and has broad applications across multiple industries. It may make a flashy press-release for your organization today but won’t make much difference in your company financials (long term or short term). Identify a bleeding area in your organization and see if an AI application can stop bleeding (we call this “hair on fire” problems). Making an unprofitable business break-even or slightly profitable is more appealing story than making a profitable business a little more profitable. Once you show early success, then you should build a portfolio of different problems.

Diversity in Your Team: As a true inventor, Carl had doubts. He lived in a different world and continuously developed the product. Bertha’s involvement was crucial. She had stronger resilience and conviction. She wanted to see her husband becoming successful. She perhaps had a clearer vision of the impact of the invention. Carl and Bertha complemented each other, and that is undoubtedly a big reason for the ultimate success. The key to diversity is: you need to act on it from day 1. Law of large number says: it will take a monumental (nearly impossible) effort to bring diversity to a not-so-diverse team once it reaches a critical mass.

Becoming AI-driven is not just a technology problem. It involves technology, people, and operations. A team of people with different background, mindsets, strengths, and weaknesses is instrumental to your success in the AI-driven organization initiatives.

After the 2008 Financial Crisis, Peter Hancock founded AIG Science —to bring quantitative and evidence-based practices across AIG and followed this principle extremely well. This enterprise-level group had not only many best-in-class data scientists, but also top economists, and rock-star management consultants.

“this multidisciplinary approach is essential to go beyond merely generating new insights from data but also to systematically enhance individual human judgment in real business contexts.” — How AIG Moved Toward Evidence-Based Decision Making, Harvard Business Review, 2014

We came a long way since then. Technology allowed us to build what we call: Automated Machine Learning. A centralized data science team e.g. AIG Science perhaps doesn’t make much sense in 2018 but the diversity that AIG Science exhibited has become more relevant. If we want to bring AI and evidence-based decision making across the organization — we need to include everyone in the process. Automated Machine Learning does that. As an Executive you should shift your focus from “data scientist” to “data science” (as famously quoted in Moneyball: “your goal shouldn’t be to buy players, your goal should be to buy wins.”).

“The journey of a thousand miles begins with one step.” — Lao Tzu

Bertha Benz changed the course of humanity entirely with her “short trip.” Following her 172-mile round-trip, we have traveled zillions of miles with no plan to stop.

It is the time that you start yours. You will be successful — when everyone in your organization is improving their decisions and judgements, thanks to the AI generated predictions.

— — — — — — — —

I thank two of my colleagues: Mandi Moon and Ashley Smith for their valuable edits of this article (at lightening speed).

It makes perfect sense to dedicate this article to my economist wife, Aparna Sengupta. Aparna supported me in every adventure that I started off. I suspect it is that curiosity.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Satadru Sengupta

Satadru Sengupta

Founder @ Halos Insurance. Ran insurance @ DataRobot. Ex AIG, Liberty Mutual, Deloitte. Believes — technology, when added with humanity, creates a better world!