Because AI Is The New Electricity
There was a time long ago when the clothes we wear, the commodities we use were created end to end by a single person. The evolution of manufacturing industries, keeping in pace with the advancements, has caused a tectonic shift in the kind of roles and responsibility of an involved individual.
Technological development was once upon a time plagued by massive developmental cycles and waterfalls ,only to be replaced by the never ending challenging lean methodologies of development. The quest to build Products for targeted user needs and using experimentation and iteration as the primary tools of strategy for achieving perfection gave rise to the massive creation and popularization of Product Management roles.
The history of Product Management, with origins in the FMCG, post adoption into tech is still being very much written. The ink with which the first Product Management roles were written is not yet dry, and the role is very much evolving.
Artificial Intelligence is yet another area which promises to grow into a behemoth and change the face of Tech. As Andrew Ng says “AI is the new Electricity”.
While the Tech world aims to achieve human-like and general machine intelligence, we have yet not surpassed the phase where AI can be leveraged and utilized for task based jobs.
The field of AI is still a mist covered forest where things remain undiscovered and mysterious and the direction is unclear and plagued by uncertainty.
An AI Product Manager role, unsurprisingly, manages to inherit the uncertainty, and the lack of a written in stone methodologies of its constituents. The tech world has witnessed a significant rise in the number of companies hiring for the role of an AI Product Manager, and yet the expectations of the companies and the candidates are often disparate in nature. The article aims to bring to the forefront the common myths of the populous around AI Product Management.
My Story as an AI Product Manager
Straight out of my MBA, I decided to take up a Product role at a small startup which could not have been a more challenging and yet interesting role. Creating the Product from almost scratch, naming the product for the first time (the naive name of Smart Surveillance) and helping the product take the baby steps of Business Model definition and pricing strategy formulation were just a few of the privileges I received. The Product assured safety and security to its users by proactively delivering insights from its CCTV or other camera feeds using Artificial Intelligence. I was nothing but more than ready for the challenge of marrying the Business to the highly technical and brilliant-mind dominated AI Development teams. Some roller coaster that was. My next stint was managing the AI-NLP engine powering a chatbot used for customer interaction in the BFSI segment. My role spanned from talking to the customers to monitoring the users to troubleshooting (why does the classification engine not deliver the 80% accuracy as promised, or why is this one query not working) to interfacing with the AI and Engineering teams on what to build next. My third and current role also involved AI, but now from across the table. While I manage a HCM Document Management Product, my main aim is to help Human Resource users manage the millions and billions of documents in the organization and tackle all the challenges that come with it. Quite a lot of the problems revolve around finding the right document, and getting answers without having to read lengthy 100-pagers. A lot of the current problems are suitable candidates for Artificial Intelligence to show its magic. But a lot about myself, let’s continue on the intent with which the article was written — to share my experiences, my wins and failures and more importantly bust the myths and I truly believe that this could help demist and extend understanding and journey of being a Product Manager or an AI product manager.
AI Product Manager Has A “Written In Stone Job Description”
Most Product Manager JDs are alike. Apart from the number of years of experience desired by the organization (which is in fact usually indicative of the kind of experience they are looking for) most organizations are looking for Problem Solving, Stakeholder Management, and Product thinking skills. But to our surprise, most AI product manager JDs almost always ask for the above mentioned set of skills. Different organizations, by virtue of the way they are organized and function, require its AI product Managers to perform a different set of roles and possess knowledge of a varying areas. Some organizations might rely on its AI product Managers to shoulder the tech-heavy interactions while there are others which require clear and deep understanding of where in Product could AI be leveraged. More often than not, the individuals at these two ends of the spectrum do not share the same career graph, an understanding that most often organizations ignore when drafting their job descriptions. Pro-Tip: Dig deeper into the company, team, and Products to understand what is expected of an AI product Manager in the organization.
AI Product Manager Needs To Be A Data Scientist
The job and responsibility of every Product Manager, while very similar, involves different weights to different responsibilities. Some Product Managers focus on internal users trying to make their lives easier, then there are those who need to be closer to their Engineering teams trying to get the product vision executed and then there are others who need to man/woman the customer side being the bridge and provide the key value of derive customer insights in order to decide what to build next. Most product managers we talk to say they do everything, and yes they do but the focus on one can be significantly more than their focus on the other areas. Product Management jobs lie across a wide spectrum and people automatically gravitate towards the role they like slowly cementing their preferences and responsibilities. The role depends as much on the organization, the culture as on an individual preference.
AI Product Management shares the attribute of being spread across a wide spectrum. An AI product manager could be someone who knows all the algorithms, what they are used for and how they work, or s/he could be someone who focuses on user problems, is sensitive to and knowledgeable of which of these problems can or cannot be solved with AI and builds out a roadmap accordingly. An AI Product Manager can be someone with extensive Data scientist experience or someone with extensive business experience. The demands of proportion for Technical and business expertise vary across organizations.
Its Solely The Job Of AI Product Managers To Build Out Roadmaps
Regular Product Management enjoys the privilege and the burden of knowing what their customers need and exactly what to build to ascertain customer happiness. More often than not these features do not battle the war of technical feasibility. Not very often are situations encountered where entire features are impossible to build, but they are more likely to lie in the “Not this way, the other way” spectrum. In such cases, Product Managers can to a large extent own the “WHATs” An AI product manager needs to strike a very delicate and exactly right balance between he customers and the AI teams. Not everything that the customer wants can be built, and not everything that the AI developers think is brilliant is right for the customers. The right balance usually includes maintaining full transparency with the technical teams about the user problems that exist and needs to be focused on, and having the roadmap always informed with the combination of user knowledge, AI research advances and sometimes the whims and fancies of the paying business executives.
AI Products Are Perfect
As with the job of a Product Manager, the job of an AI product Manager also includes setting correct customer expectations. More often than not customers do not know what they need (they might know what they want), and they definitely do not know what is remotely possible in the realm of AI. Clients do have this bad habit of quoting accuracy in their business deals and acceptance of products. But the very fact that there was a need for AI to solve a problem or a set of problems is because of the failure of regular non-AI based methodologies to achieve the goal. It usually falls upon the AI product managers to transparently communicate the accuracy and the performance metrics to the customers. Solving a big problem partially is always better than no solution at all. If millions of people are reaching out to you everyday, having an army of customer service agents is neither profitable nor judicious use of resources. The variations in conve