Every day, it feels like founders come up with a new adjective to describe the potential of artificial intelligence. And they’re not wrong—AI has truly demonstrated its transformative capabilities over the past few years.
As a result, everyone wants to add an element of AI in their applications, and it’s now making its way into products used across generations. Whether it’s an email client, messaging apps, or a suite of AI features that show up after updating smartphones.
But is it fair not to let users disable these additions or force them to go through a cumbersome process of hunting for buttons buried deep within menus to turn them off?
Such has been the case with certain applications, and several users have expressed their frustrations on social media. While some have been unable to disable AI features, others are struggling to find the settings to do so, on products offered by companies like Meta, Microsoft, Google, Apple, xAI, and so on.
A controversial example is Snapchat, which introduced the ‘My AI’ chatbot. Despite backlash, the company only allowed users with a Plus subscription to disable it, sparking further frustration among non-subscribers.
While most of these products let users opt out of sharing their data for the purposes of training, some don’t want these features to clutter, or interfere with their usage experience.
To understand the rationale behind such strategies, AIM spoke to Karan Peri, an independent product advisor, with over a decade of experience in product management at companies like Microsoft, Coinbase and Amazon, among others.
‘Trying to Please Everyone Will Only Lead to Failure’
When the company decides to introduce a feature, it is almost always based on an A/B test, which involves testing two product versions among the users.
The one that rates better on an aggregate makes it to the final version. Thus, in this process, the preference of individual users isn’t considered, only the majority’s.
Peri stated that if a feature does not provide value for a long time, good product teams will either iterate or turn it off, especially if the backlash is detrimental to the engagement of the product.
However, when companies have to consider the feedback of a minority section, and end up adding more knobs and controls to the feature, changes to the code base can increase the maintenance costs.
“If you go on a path believing that you want to please everybody, you will fail. That means you’re dropping your quality to the least common multiple. You keep dropping till it pleases everybody,” Peri added.
For instance, a 75-year-old user of a popular messaging app is frustrated by the AI features she can’t disable. She falls into the third group of people who dislike the feature, alongside power users in the first group who love it, and the second group consisting of those who don’t care about the feature.
“But the first two buckets are way larger than the third bucket, so the company doesn’t care,” Peri explained.
Nevertheless, if the messaging app occupies almost a monopolistic position in the industry, its competitive position is advantageous. “If not for that app, where would the 75-year-old woman go?” he questioned. “She may not go to another app because her grandchildren, and other family members aren’t there. Whether she likes it or not, she is the product has her hooked.”
Hence, certain companies will continue to retain these features.
Moreover, certain features may not appeal to people initially, considering the cognitively heavy task of understanding how it works. However, users eventually end up embracing it. Similarly, companies often aim to integrate these features within the fabric of the product, like Netflix, for example, where the AI-enabled recommendation engine has become a crucial factor in the user experience.
For example, several users complain of the inability to disable Meta’s AI features on WhatsApp, Facebook, and Instagram. However, a report from The Information revealed that Meta AI daily usage is heaviest on WhatsApp and Facebook. Meta also revealed that the chatbot had about 185 million weekly users of Meta AI across their products, as of last year.
Furthermore, despite Snapchat’s questionable choice of not letting users disable ‘My AI’ chatbot for free, the app has seen nothing but consistent growth in the number of active users over the years.
That said, strategies vary across companies. For example, Apple Intelligence provides an option to turn the features off.
“If you buy a new iPhone, which is super expensive, you must ensure that the phone is liked. If not, then the entire line will get discredited,” said Peri, indicating that the scenario takes a turn when there is hardware, and money is involved.
However, when Apple came under fire for its AI-enabled notifications features hallucinating and reporting fake news, the company eventually halted the feature.
Are Companies Making a Good Case for Value Addition?
“People are thinking about turning an AI feature off because they do not know how to get value from it,” Peri pointed out.
“If it was valuable and people understood what it was doing, maybe lesser people would have asked, ‘How do I turn it off?’ You want to turn it off because it is useless,” he added, indicating that people on forums wouldn’t want to ask how to turn it off if it added good value.
Several factors contribute to this scenario. The company likely did not instill enough thought into the launch, choosing to release it and observe user behavior without explaining how it works. Moreover, the instructions may not have been clear enough for users to understand the feature’s purpose and functionality.
However, Peri said that good product teams often implement a feature in a way where education isn’t needed.
In conclusion, it boils down to a single fact—Not everybody can be pleased. The goal is to please as many people as possible.
