Ground rules for applying artificial intelligence to product management
First of all, I want to thank Kobe Bryant (he died yesterday in a helicopter crash) for being such a huge inspiration. Rest In Peace Kobe and the rest of the crew!
The hype around artificial intelligence (AI) and machine learning have led to lots of jargon so that this compelling technique has become more challenging to understand. So how do product managers can leverage AI? What can they do? How could they think about opportunities which AI introduces?
The goal of this post: In this post, I am going to explore what are the strategies which product managers can develop in building AI-powered products.
Table of content:
- Product managers should focus on problem-solving first and AI can help a lot.
- What are the three levels which product managers need to consider?
- AI will empower humans rather than replace their jobs, and I believe product managers can accelerate this change.
- Artificial Intelligence can make an impact on product management in three different ways (Uber, Dell, Zappos, and Gmail examples)
Most products require decision-making based on data, though the method for making these decisions can diversify. For instance, machines or humans can make decisions, and the data behind these decisions can be static or dynamic. A focus on decision-making removes away the complexities of specific methodologies or the noise of industry jargon. This more general definition helps product managers to be more attentive to the challenge space – it removes distractions that can lead product managers to think about solutions too early in the product development process.
Product managers must fully understand the problem so they can adequately define specifications and align the team of engineers, researchers, programmers accurately (here is the post "How to staff an AI team")
What are the three levels which product managers need to consider?
Let's start quickly with what product managers' responsibilities are:
- managing product journey from concept, to design, through prototyping, testing, forecasting, scaling, promotion, support, and finally, landing the end of the product life.
- delivering product roadmap which is mostly about setting up a vision, align the vision with the organization strategy, break down the vision to strategic elements which can be converted to task later
- meeting or pivoting growth objectives, including market share, revenue, profit, and return on investment for all the channels/categories of business and key customers.
So, where is artificial intelligence in all of that? As I mentioned above, if we look at this as the problem-solving machine, it will bring a lot to the product manager's job. But first, please let me introduce you to three levels of product managers' journey.
Product managers need to build and execute with the vision in their minds (roadmap requires it). Then, the strategy comes and asks for attention. A strategy is about doing the right things. Nokia did the right job and pivoted from the rubber company to a cell phone company, but Microsoft did not do the right thing (tactic) by acquiring Nokia and ignoring iPhones keyless keyboard.
Product managers need to operate 24/7 on the following levels, but AI sits only on one of them:
- The vision which is a True North
- Strategy - what is the right thing I need to do
- Tactic - how to deliver the right thing (project plan level)
AI sits on the tactic level. It is designed by its nature to solve specific, well defined, data backed up, problems. Of course, to bring AI as the problem solver requires a strategic decision, but it will not embrace the strategy without an excellent tactic. Shipping AI without an end goal usually brings no value to end-users.
The Utah-based company, Hire-Vue, began using AI in 2014 as a way to help companies sort through video interviews. The team defined a great problem to solve, how to select candidates leveraging video data. HireVue believes it can be helpful for processing a huge number of people through the interview process quickly and reviewing the video in a consistent way.
Another example is AirBnB. The company has developed technology that looks at guests’ online “personalities” when they book a stay to score the risk of them trashing a host’s home. AirBnB technology searches for fake accounts, users who publish alcohol or sex video, etc.
These two examples show a clear problem, on a tactic level.
If you want to subscribe to The Innovation Letter and get the next article into the mailbox, please navigate here and Subscribe.
AI will empower humans rather than replace their jobs, and I believe product managers can accelerate this change.
The widespread discussion on matters like automation has raised some interesting issues about the future of work, and finally, the narrative around how AI can help humans have become forgotten.
A typical example that illustrates this point is self-driving cars, the car-sharing industry, as well as outsourcing.
Within the automotive industry, autonomous capabilities for vehicles are classified into five categories (only the highest class consists of fully autonomous cars in every condition).
The car-sharing industry can lower the demand for cars, which will influence the job market. At the same time, the cybersecurity industry will create numerous jobs because those shared cars need to be safe and reliable.
In outsourcing, many tasks can be automated, for sure. The only important thing is that automation requires a clear, well-labeled data set that will teach the AI. Human beings need to teach AI how to automate a task. It creates space for people's development as these tasks are not the ones that help people to get better in their jobs; instead, they drag people away from becoming better professionals.
Product managers need to recognize that high-performing AI ecosystems need a sizable dataset to get started, and datasets need to be well-structured high volume, and machine-friendly. Ideally, the dataset should also have well-defined notions of success and failure, where past outcomes are predictive of future results. The problems require this, and if the data is weak, AI will not help, at least nowadays. There is another critical factor, and only people can understand the real world and transform the observations to structured rules. American computer architect Frederick Brooks communicates in the book The Mythical Man-Month that, in spite of the progress brought by AI, it does not have the human faculty of understanding, which makes it incompetent in writing software.
So. AI needs people to solve problems, and this is excellent news!
Artificial Intelligence can make an impact on product management in three different ways.
AI can easily automate operational processes, which are time-wasters or manual activities today. Also, products with comprehensive behavioral tracking can generate a dataset that empowers teams to spot and define processes which are automation ready. For instance, customer touchpoints and communications can be optimized based on data to increase conversion or reduce churn. One of the most common examples is a repeatable and standardized questions list, called FAQ. The list is simple for AI to understand; the best answers, output, are known already. Then, AI serves as an advanced business intelligence engine that accelerates productivity and effectiveness for clients.
AI can significantly improve the user experience of products and services. Please find four examples of how Uber, Dell, Zappos, and Google have been able to leverage AI to create higher-value experiences for their customers:
|What has an AI-powered |
|Dell||Laptop||The company embedded AI-powered|
optimizer into the new Laptop Latitude
9150. AI-enabled the machine to learn
how a user actually uses the laptop, so
Dell can optimize battery cycles, software
algorithms and audio. It will give
professional users a power to adjust
Dell to specific requirements.
|Uber||Mobility||Uber leverages AI to conduct better|
matchmaking between drivers and haulers
as well as calculate prices and ETA more
|Zappos||Search engine||Zappos's team built better working|
term recognition software into the
eCommerce giant search engine. The model
imports historical data and immediately
picks up on new keywords that are
introduced as soon as fashion trends start
to pick up speed.
|Gmail||Through the use of machine |
learning algorithms, Gmail successfully
filters 99.9% of spam.
To sum up, here are six ground rules which help product managers consider integrating AI into their product development work:
- Define AI as the more broad term, and it will help you to focus on problem-solving rather than the end solution.
- AI is a tactic that should be leveraged to solve well-defined problems, not a strategy, although it should be aligned with one.
- AI can empower humans, creates opportunities, rather than replace people, and eliminate jobs.
- AI drives impact on the product roadmap in three ways: automate processes, improve product experiences, and create new products.
- AI needs people as it can't understand the real world and transform the observations to structured data sets.
- AI product managers need to orchestrated a team of engineers, AI researchers, ML engineers to one squad, which will undertake the problem-solving path.
WOULD YOU LIKE AREK TO WORK WITH YOUR ORGANIZATION?
If you are interested in working with Arek Skuza for a business project, workshop, speech or presentation, please fill out the form here.