Impact of AI on Product Management [Malte Scholz]
What is the impact of AI on product management?
Let’s get right into it!
- According to a study by Airfocus, 92% of product managers believe AI will have a huge impact on their work in the future. 70% of PMs are concerned AI might take their jobs whereas 21% feel they don’t have adequate skills to use AI effectively.
- There’s no data indicating that AI might steal your job.
- However, there seems to be little acknowledgment of the harm that AI could cause, for example, through interference in democratic elections.
- There are two basic kinds of AI models. You can either use AI models trained on generic data, like OpenAI, or custom models trained on your own data.
- If you integrate an existing model into your product, like most SaaS companies, it won’t be enough to outperform your competitors.
- Training your own model requires access to data and technical resources but could be a true differentiator in the market.
- AI is excellent for data analysis, pattern recognition, and automation. However, it lacks emotional intelligence or contextual understanding.
- Product management is all about solving problems and in this respect, AI can be of immense help.
- Automation frees up PMs’ time so that they can focus on activities that make the biggest difference, like product strategy development.
- As it’s great at data analysis and pattern recognition, AI also helps PMs make better-informed decisions.
- When integrated well, AI can enhance the product experience for customers.
- Start implementing AI by identifying a problem to solve. Only then choose the right (AI) tool to tackle it.
- Next, make sure that delivering the AI solution at scale is feasible.
- Finally, consider the impact on users, as not all will be able to benefit from it.
- Userpilot uses AI to power its writing assistant and localization features. Book the demo to learn more!
What’s the buzz about AI?
Ever heard of the 4th Industrial Revolution? Or Industry 4.0? Maybe 4IR?
In short, this is the shift in how we live, work, and communicate that we’re experiencing right now, brought about by the omnipresent digitization.
AI is one of the driving forces of 4IR. According to World Economic Forum data, the number of jobs requiring the use of AI has increased by 450% since 2013.
Product management is also affected by the trend.
92% of the PMs who took part in the 2022 study on the impact of AI on product management believed that AI is going to have a significant impact on product management.
However, this enthusiasm is accompanied by concerns.
A significant 70% of product managers express fears of AI potentially replacing their jobs, while 21% of PMs feel they lack the knowledge and skills required to leverage AI effectively in their roles.
The Impact of AI on Product Management by Airfocus.
The two faces of AI: Overhyped and underhyped
The views on the impact of AI seem to be very polarized.
On the one hand, we have the overhyped fear that AI may steal human jobs. However, record-low unemployment rates suggest that this fear is not justified.
On the other hand, there is some harm that AI can cause.
The fact is that AI’s power can be exploited to wreak havoc and destruction, for example, in the form of terrorist attacks or interference in democratic election processes.
This raises concerns that we should perhaps look more closely at its regulation.
Understanding AI models
AI models are like learning kids: they observe, learn, and act. In other ways, you provide them with input, and they process and internalize it, and use it to provide output.
There are two ways to implement AI models.
The easiest way is to use the available API solutions like OpenAI. Such models are trained on generic data, and so is their output.
To achieve customized output specific to your field, you need to train the model yourself – based on your own data. This alternative, however, requires considerable investment and expertise, so may not be practical for applications with limited use.
How to train AI models?
The training process can take different forms.
Supervised learning is like teaching with an answer key. You provide the AI with labeled data that contains input-output pairs.
Unsupervised learning, on the other hand, allows AI models to explore data without labeled guidance. It is the “self-discovery” approach, where AI autonomously identifies patterns, clusters, or structures within the data.
Finally, reinforcement learning is like teaching the dog tricks by rewarding them with treats. Positive and negative feedback shows the AI what kind of output you expect and helps make better decisions in the future.
AI business models
Currently, companies tend to integrate AI into their products in 2 main ways.
First, we have the ‘table stakes’ way.
That’s when you take a ready AI model, like the OpenAI one, and embed it in the product to enhance its functionality and increase user efficiency.
Think of tools like Notion embedding AI writing assistant or feedback tools like Akkio leveraging AI to interpret user sentiment and find patterns.
As customers expect modern AI tools to have AI functionality, it will soon become the norm. However, it won’t be enough to differentiate your product from the competition.
Alternatively, you can use the gold mine approach. This requires building your own model on your own data.
A good example is Grammarly, which uses its own data sets to train its proofreading and writing tools.
This is very time-consuming and expensive but also difficult to copy for competitors. Consequently, it could easily be your key differentiator.
AI’s sweet spots
There are a bunch of things that AI is pretty damn good at.
Let’s start with data analysis.
Another area where AI excels is pattern recognition.
Finally, let’s not forget about automation.
AI can streamline and automate numerous tasks and processes, reducing manual workloads and human error.
Want a good example?
Check out how Miro uses AI to help users analyze content and automate the production of visuals. For example, you could use it to easily turn the outcomes of a feedback or brainstorming session into a slide deck.
What AI can’t do (yet)
Despite its strengths, there are distinct limitations to what AI can achieve.
Take emotional intelligence as an example.
AI, as powerful as it may be, still falls short when it comes to emulating human empathy and understanding. Human emotions are simply too complex and nuanced for AI to understand at the moment.
Funnily enough, people are not always great at that either, but that’s a different story.
Anyway, we’re not likely to be using digital therapists any soon.
Other areas where AI lacks are contextual understanding, complex decision-making, and creative innovation.
The role of AI in product management
Considering its strengths and weaknesses, what role could AI play in product management?
AI will definitely not replace product managers in the foreseeable future. It simply doesn’t have the people skills or creativity that are so essential in product management.
However, AI is a tool that offers immense help to PMs. Product management is very much about problem-solving, and AI can help product professionals solve complex issues quickly and efficiently.
The trick is to know what kind of problems can be solved with AI and what tools to use to solve them. That’s what PMs must focus on to not fall behind. As Richard Baldwin puts it, ‘AI won’t take your job, but a person with AI will.’
Using AI in product management
Instead of looking at AI as a threat, PMs should view it as their sidekick: not able to do their job independently, but still incredibly helpful and valuable.
How exactly can it help you?
For example, by automating routine and repetitive tasks like report generation, or performance tracking.
Second, AI makes PMs faster and more efficient.
For example, they can use it to analyze quantitative data or gain deeper insights into how customers perceive their products from reviews, chat logs, or social media interactions.
As a result, product managers can make informed decisions based on data-backed evidence rather than intuition alone.
Finally, AI enables PMs to delight customers by delivering a better product experience. When implemented well in your product, AI technologies can help customers achieve their goals.
AI in Airfocus
Airfocus is a product management platform.
One of its features is the feedback and insights workspace. It enables product teams to consolidate customer feedback from different channels in one place.
That’s where Airfocus has integrated AI technology.
Here’s how it works:
Some of the options that it offers include:
- Analyze sentiment
- Explain this
- Find action items
All you have to do is choose the pieces of feedback you want to analyze and choose the desired option, and AI Assist will do the rest for you in no time.
Thanks to AI Assist, product managers can easily analyze huge quantities of customer feedback and identify patterns.
Challenges and considerations before implementing AI
Before you start implementing AI, either to optimize your processes or to improve the product, consider a few things.
First, choose AI to solve specific problems. If you can identify a specific area where AI can make your life easier or enhance the product, go for it.
However, don’t use AI for its own sake. Refrain from chasing the new shiny object even if everybody else around is doing so.
Next, assess the feasibility of implementing AI solutions.
Do you have the necessary data and financial or human resources to construct a proprietary model? Is it worth doing it or are you better off using a publicly available API? Or perhaps, should you choose a middle way, where you train existing models on your own data?
Finally, think about the impact on the user and their skills. Don’t assume that all users will be able to understand AI outputs – that’s not a given. Not all user personas will be able to grasp AI, and it won’t necessarily benefit all of them.
How does Userpilot leverage AI?
Userpilot is a product growth platform.
It enables product teams to track user behavior inside web apps and collect feedback via in-app surveys. Thanks to the engagement layer, you can create interactive onboarding experiences, provide in-app guidance, and drive customer engagement.
Userpilot offers 2 AI-powered features.
The AI writing assistant is one.
You can use it to create your microcopy from scratch and tweak your onboarding flows by adding relevant text, summarizing it, or extending it. Not going to mention fixing spelling or grammar mistakes – it’s a given.
The other one is localization.
In addition, there are two more AI features in the pipeline.
Users will soon be able to use AI to analyze qualitative customer feedback at scale. This will help product, marketing, and customer success teams to quickly identify patterns and key themes in user feedback.
The predictive analytics feature is even more exciting. It will enable teams to extract insights from historical user behavior data to make accurate forecasts.
For example, you will be able to predict your retention and churn rates.
- AI is real in product management: it’s a true game changer, not just hype.
- PMs, be aware of AI limitations: it’s powerful but not emotionally intelligent.
- AI is your ally: it enhances PM roles, but it doesn’t replace them.
- Start with the problem: use AI to solve real issues, not because it’s trendy.
- Skill-up: basic AI knowledge helps PMs and end-users alike.
If you’d like to learn more about Userpilot’s AI features, book the demo!