Skip to main content

Author: markozivanovic

How AI Personalization Fuels Groupthink and Uniformity

No, this isn’t another article crapping all over AI and babbling about our AI overlords replacing us at the top of the food chain. It’s a critical take on the ever-rising trend of pushing AI-powered suggestion and recommendation mechanisms in every service and tool we use.

I read Slack’s privacy principles on search, learning, and artificial intelligence a few days ago. Like many users, I wasn’t delighted to see how Saleforces’ Slack will use private conversations from its clients to train its AI. There’s much backlash about it online, and everyone and their mother already shared their opinion about it somewhere. I see no point in adding to that pile, but I find another part of that text as concerning.

Right now, I want to focus on how the data will be used. Please note that Slack is not the only one (or the first, for that matter) doing this. This was only the trigger that made me think again about how the bubbles we live in will become even smaller. Slack is the primary communication medium for many businesses worldwide. Let me be idealistic for a moment and say that all those businesses want to create value. To create value, things such as creativity, new ideas, original thoughts, and something you want to have more of, right?

Here are some ways how Slack will use their customer’s data to “make your life easier”:


  1. Autocomplete: Slack might make suggestions to complete search queries or other text– for example autocompleting the phrase “Customer Support” after a user types the first several letters of this phrase.
  2. Emoji Suggestion: Slack might suggest emoji reactions to messages using the content and sentiment of the message, the historic usage of the emoji, and the frequency of use of the emoji on the team in various contexts.
  3. Search Results: Our search machine learning models help users find what they’re seeking by identifying the right results for a particular query. We do this based on historical search results and previous engagements (…)

At first glance, these features seem harmless, even helpful. They save time, reduce friction, and enhance user experience. However, beneath the surface lies a more troubling consequence: the potential for these features to stifle creativity and reinforce groupthink.

Consider the autocomplete function. By suggesting common completions based on past data, Slack’s AI could inadvertently discourage users from thinking outside the box. If the AI continually nudges users toward conventional phrases and ideas, it might limit the expression of novel thoughts. Over time, this can create a feedback loop where the most common ideas become even more dominant.

Emoji suggestions present a similar issue. If certain emojis are suggested more frequently because they have been used more often in the past, it could lead to a homogenization of emotional expression. This might seem trivial, but it’s one more thing that makes the whole communication experience uniform, predictable, and boring.

Search results, too, are influenced by past behaviors. If AI algorithms prioritize results based on what has been frequently searched for and engaged with, they can create a self-reinforcing loop that amplifies popular ideas and suppresses less common ones. If everyone is guided to the same set of information, the range of ideas and solutions considered will narrow.

In essence, AI personalization in tools can lead to an echo chamber. It can lead to uniformity and groupthink.

There’s a popular saying, “Always use the right for the job.” Maybe we should add a part that goes, “if it’s not dangerous in the long run.”

The roaming office: Don’t be *that* remote worker

I probably (most definitely) sound like an old man yelling at a cloud, but if your work consists of many online meetings, try to plan so you have a stable internet connection and quiet surroundings during the meetings.

There are exceptions. You’re absolved if you often travel for work and spend most of your time on the road by design! It’s totally fine if you’re heading for a conference or a client/stakeholder meeting, and the only way to unblock the others is by tuning in from a busy train or café.

I have a bone to pick with the other group. The ones who forgot they’re conducting business meetings, so they treat them like Facetime with friends.

I’m writing this to raise awareness. I know it’s easy to lull yourself into thinking it’s okay to do what you’re doing. You were doing some chores one day and wouldn’t make it home in time. So, you did one meeting from the road. No one said anything, so you did another the next Wednesday. Fast-forward a few weeks, and you’re running a two-hour workshop from a busy bar.

OK, I’m also writing this to vent a bit, but raising awareness is a big part of the reason. I promise! At this point, I’ve really seen a lot since 2020.

  • I did (well, tried to do) a job interview with a candidate who couldn’t hear me properly because they were in a park surrounded by a bunch of people
  • I sat through a meeting where a person tried to share their screen who knows how and do a presentation while walking
  • I had a meeting with someone who was on a beach, with people playing volleyball in the background

I support giving people as much flexibility as possible while doing the work. I’m trying to get across the point that we shouldn’t forget there’s a line between work and play, and that line should be a bit clearer when one’s actions impact other people’s time.

If you plan to work from a place other than your home, coworking space, or any other space where you have proper working conditions, make sure in advance to ensure:

  • The Internet connection is stable enough for the type of work you’re doing (people usually don’t ask that; they just see that the hotel/motel has a working connection and then realize it’s not sufficient after the fact)
  • The place you’ll be working from is quiet, and you can have a private space if needed, depending on the nature of your work
  • You will not be disturbed or must leave/change the space during the meetings

If you work in a place that’s supportive of remote work and flexible about how and when you work, don’t take it for granted. Don’t be that person because of whom we all can’t have nice things.

I’ll take it slow today

It’s one of the trends that came with WFH being more accepted (hello, COVID, my (old) friend) in the workplace.

People will often show up online and say something like, “I’ll take it slow today,” “I’m not 100%, but I’ll try to do something,” or they will just put a battery icon with almost depleted energy as their status, followed by some text along the lines I mentioned previously. It’s supposed to mean that even though the person is not feeling great, they’re not feeling that bad, so they can’t do their work. The only caveat is – that no one knows for sure what it means.

Have you noticed the trend? Are you one of the “I’ll take it slow today” people? Don’t worry; I’m not judging. I want to tell you why this may not be a great idea. Even though it’s probably used with the greatest intentions, the whole “taking it slowly” thing is disruptive for both sides—the one taking it slowly and all the others who usually interact with that person.

Here are some reasons against it:

  1. Expectations are not clear. What does your status mean? Should people not contact you or set new meetings with you? How long will it take you to respond to the messages? Can your coworkers still count on your pre-agreed workload being finished while you’re taking it slowly?
  2. You’re creating an unhealthy culture. By sending out a message that you’re working while not feeling well, you’re normalizing working while being sick. It’s even worse if you hold a more senior position in the company. Remember the “lead by example” motto.
  3. The evaluation period is not divided into slow and regular days. There’s no “working slowly” option in the management’s calendar. No one will adjust your evaluation to account for you not being 100%. Even worse, you risk making mistakes when you’re not feeling well. It’s a lose-lose situation for your career progression.
  4. You make others feel guilty. People see your status, and they evaluate whether their request is important enough to bother you while you’re not feeling well.

If you’re not feeling well enough to work, don’t. Focus on recovery and return to work when you can handle your tasks as you would regularly. If you decide to work even though you’re under the weather, clarify your availability and scope of work. If you choose to follow that path, here’s an example of how to handle that communication.

“Hello [team, person]. I’m not feeling that great today, but I’m OK enough to take the meeting with our stakeholders, as it’s a big pain to reschedule. I will be online from 11:30 to 13:00 to see that through and create the meeting notes. After that, I’ll take time to recover, and I’ll be offline. I’ll keep you updated on my status!”


When it comes to communication in the workplace, try to be as precise as possible. Also, take care of your health, mental and physical.

Some productivity metrics are evil

You have the company, the vision, the product, and the plan to improve it and make it more feature rich. You also have the stakeholders, who are anxious to see those new features delivered. If you’re responsible for that delivery, now you also have a problem!

You don’t want to overpromise, nor do you want to underpromise. One thing is sure: you must promise something; commitment is not optional.

This is where the problem starts. If you overpromise, you’ll probably miss every deadline, putting immense pressure on the people working on the deliverables. If you underpromise, you’re not realizing the full potential of your product teams, and you’re wasting resources.

So, you do the sensible thing. You tell the product teams to plan their sprint (or any other unit you use to measure an interval of work), committing only to the number of deliverables they are highly confident of guaranteeing. Teams will do their due diligence and ensure the number of deliverables is optimal, with maximum utilization of the team’s resources and a “quality first” mindset.

And they lived happily ever after—the end.

Well… not exactly. Unfortunately, the reality is a bit different. The story above is just one side of the medal. The other side is how you run the company and evaluate the performance of the teams and their members.

To assess performance, you need some metrics. And this is where it can get tricky. Metrics are problematic because if you’re not careful about how you set those, you can end up incentivizing rather unhealthy and unwanted behavior.

Let me use a simple example from our story at the beginning of this post to discuss “deliverables.”

Let’s say you have two teams, a blue and a yellow team. You decided to introduce a metric that shows whether the teams responsible can achieve a pre-agreed number of deliveries per sprint. As delivery is a pretty ambiguous term, you agreed to talk to both teams before each sprint, discuss the complexity of the deliverables, and see if you can decide on how many deliverables they can commit to.

After ten sprints, you look at the metrics. A blue team committed each sprint to four deliverables, achieving their goal every sprint. On the other hand, the yellow team committed each sprint to three deliverables, but they failed to meet their goal three times out of ten.

Your metrics tell you that the blue team should be rewarded, and the yellow team should have their performance evaluated. I’m oversimplifying this situation, and any sensible manager wouldn’t reward or reprimand the team based on one metric. Still, organizations often have many loosely defined metrics, which can all lead to questionable actions.

While it might be true that the blue team overperforms the yellow team, you can’t be sure. The metric used in the example leaves a lot of possibilities (and some questionable behavior) open. Here are some examples:

  1. The blue team is better at overselling the complexity of the deliverables, lowering the actual output they’ll commit to. The yellow team might be better at sales but not necessarily at delivery.
  2. The blue team decides to overlook quality (testing, covering edge cases, cleaning technical debt, etc.), optimizing for shipping as many features as possible but endangering the product long-term.
  3. The blue team marks some product requirements as not feasible during the discovery/refinement sessions to avoid dealing with potentially uncertain work (albeit possibly yielding high rewards for the company)
  4. The yellow team could’ve possibly delivered multiples of what the yellow team achieved, always trying to push themselves more each sprint. They optimized for value and set ambitious goals to advance the product further.

My examples describe the blue team as toxic, and this might be extreme, but it illustrates the point pretty well. Anecdotally, working in several toxic environments in the past, I witnessed this behavior first-hand.

Even though it’s rarely the intention of the person responsible, some metrics they introduce to their teams can be EVIL! This is how I define an evil metric.

An “EVIL METRIC” is a performance metric that, due to its design or the context in which it is applied, can lead to behaviors contrary to the best interests of the team, product, or company.

It promotes short-term gains at the expense of long-term success, quality, and team morale. It can lead to toxic behaviors such as gaming the system, overselling, underdelivering, neglecting quality, and avoiding challenging but potentially rewarding work.

How to avoid implementing evil metrics?

Here are some questions you can ask yourself when you’re thinking about the metrics you want to use to assess the performance of your team(s):

  • What behaviors will this metric incentivize? Are there any negative behaviors it might inadvertently promote?
  • Does this metric favor quantity over quality, or does it balance both?
  • Is the metric understandable, and can team members see how their actions directly influence it?
  • How easy is it to game or manipulate this metric? What safeguards are in place to prevent this?
  • Does the metric fairly evaluate all team members, considering the diversity of roles and tasks?
  • How will this metric be used in conjunction with rewards and recognition? Will it encourage teamwork and collaboration or foster unhealthy competition?

Final thoughts

Ensure the metrics you use promote fair play and are not prone to manipulation. Discuss the metrics with the team and gather feedback. You should never use metrics that are not transparent to those who will be evaluated based on them.

Reporting on progress and communicating with stakeholders don’t have to overlap with evaluating your team’s productivity. There is a balance, but it requires a lot of work to establish. All the good things do! Happy metric setting, and don’t be evil!

Considerate Digital Communication

Considerate Digital Communication is the practice of using digital communication tools in a respectful and mindful manner that prioritizes the time, attention, and well-being of others. It aims to reduce digital noise, prevent spam-like behaviors, and promote a more efficient exchange of information.

You probably recognized several patterns of non-considerate communication in your day-to-day work. It’s a big part of why we must use “mute buttons” and suppress notifications in our instant communication tools such as Slack, MS Teams, etc.

There are many examples of abusing and misusing communication tools that make our lives a bit noisier than they should be. Maybe you’re also the culprit, and you’re not even aware of it.

Here are four guidelines for being more considerate of people’s time regarding how you communicate digitally.

Note to readers: I coined the phrase “Considerate Digital Communication” as an umbrella term for anything fitting the definition at the beginning of the article. If you have any input or ideas for more guidelines, I’d love to hear those. The best way to contact me is by email, which you can find on the homepage.

One message ought to be enough

If you send someone an email, tag them in the project management app, tag them in another channel, etc., DON’T immediately send them another message that you did so. If you created any action that resulted in the notification for the receiver, don’t double it, reminding them about it immediately. It creates noise and it’s spammy behavior.

Maybe you’re frustrated that people are not paying attention or ignoring crucial messages, but you’re not doing anyone favors by nannying your colleagues. Everyone is responsible for doing their work and following up on the work they’re responsible for. If that’s not working out, that’s a topic for another conversation (and a serious one). That behavior is never fixed by repeatedly messaging someone about the same thing.

Courteous, properly spaced reminders are, of course, fine. A good example of this would be reminding a group of meeting attendees about checking their prep work before the event that was scheduled some time ago.

Discourage the stream-of-consciousness writing

Definition: Stream of consciousness is a narrative style that tries to capture a character’s thought process realistically.

Here’s a hypothetical example of stream-of-consciousness writing about a problem John is trying to relay to Tom:

10:23 AM John: hey, Tom
10:23 AM John: two users called this morning to say that our shop is really slow
10:26 AM John: I don’t know what’s going on
10:28 AM John: can you help me find the problem
10:29 AM John: I tried to contact Sarah, but she’s offline
10:32 AM John: one user just sent me an email that everything is working fine now, but
11:13 AM John: hey, I checked the logging information for the database
11:18 AM John: I think it’s something about the regular maintenance on our provider’s side
11:21 AM John: ah, it’s not, it’s something from our side
11:23 AM John: I’m trying to think if I actually changed something, hmmm weird
11:41 AM John: ah, to hell with it; it’s all okay now

This type of communication is overly time-consuming and confusing. Let’s say that Tom opened the chat at 10:23 AM. He would stare at the screen, looking at the “user is writing…” indicator, waiting for the next message of revelation from John just to be disappointed by receiving another bit of John’s incoherent rambling.

Instead of writing a stream of short messages (generating multiple notifications in some apps), take the time to write a self-contained message. Add as much context as you can and wait for the response. If something is urgent, there are probably other (and better) ways to escalate.

Use cross-posting sparingly

There’s nothing wrong with cross-posting, but some people tend to do it excessively. Before cross-posting, think about the people that make up the channels you’re cross-posting to.

Many companies will have dedicated channels for announcements and general discussions where people are usually added on their first day of work. If you post something there, there’s little value in posting the same thing on other channels.

Cross-posts make sense if an announcement is applicable to two or more teams that lead their discussions in different channels. Companies with many cross-posts on their communication platform could benefit from better channel organization and communication strategy.

If you’re on the receiving end of many of the same messages that were cross-posted, raise that with your manager (or anyone able to help improve the processes in the company). Maybe you can investigate the root cause and hopefully make some improvements!

Avoid direct messaging people in off-hours

One of the advantages of the modern workplace is the flexibility. A lot of work can be done nowadays from anywhere, and what’s more important for this section – anytime. That means that our digital communication spaces are occupied and used 24/7.

There’s nothing wrong with that fact, but it inadvertently brings that FoMO (Fear of missing out) phenomenon with some people. Even if the workplace relentlessly works on telling its employees they shouldn’t check the work tools in the off hours, FoMO can often kick in. That’s why we can and should be considerate to each other by scheduling the messages if possible.

Most, if not all, instant messengers can schedule a message in some way. If you need to message someone directly or tag them (anything that will trigger the notification) in some channels and you’re working at a different time, schedule that message for the company’s regular working time.

By doing that, you’re also helping create a culture that respects the work-life balance and doesn’t add unnecessary pressure.