Earlier I withdrew some money from the cash machine and at one point the screen read:

"Hold on, I'm contacting the server".

The designers of the system had chosen to infer through the language used, that the machine had a personality.

My reaction as UXer was one of puzzlement/curiosity. When I asked my wife what she thought, she used the word "creepy".

Do we know anything about why this is done? Does it actually lead to generally more positive user experiences? And if so, what are the principles behind how it works?

Or is it just something that emerged from the evolving practice of UI design, and it caught on because people like the idea of it, rather than it having explicit value?

  • How else would a non-living entity refer to itself?
    – ADTC
    Commented Nov 18, 2014 at 16:39
  • 3
    It's a good thing it didn't say "hold your horses mate, I'm discussing this with my master..." Personally I find that kind of language unprofessional. I would prefer it to say "Please wait, contacting server..." There is no "I" and it conveys the message with professional language without any superfluous content.
    – ADTC
    Commented Nov 18, 2014 at 16:57
  • 1
    @ADTC I think the point is that non-living entities don't have "selves" to refer to, which is what creates the problem. Not sure what to make of your assessment of "unprofessional". If it doesn't meet professional standards, why not?
    – dennislees
    Commented Nov 18, 2014 at 17:08
  • If you speak to the receptionist of a company headquarters he or she will say "please wait, sir" not "hold on, man". That is, if maintaining a professional image is important to them. I agree that nonliving entities should not refer to themselves as "I" or "me". That's usually cheesy, and I would say definitely unprofessional in a professional setting (like your ATM).
    – ADTC
    Commented Nov 18, 2014 at 17:14
  • 3
    I think you're getting hung up on the formal/informal aspect. This isn't about the difference between "Please wait" and "Hold on". It's about the self reference aspect. You may think it's cheesy, but I'm interested in why people do it. And I'm still not sure how you maintain that an ATM saying "I'm contacting the server" is unprofessional. We're talking about the same word here, right? Unprofessional:"below or contrary to the standards expected in a particular profession." If it's unprofessional, what are the standards it fails to meet, or is contrary to?
    – dennislees
    Commented Nov 18, 2014 at 17:40

3 Answers 3


Alan Cooper makes the point IIRC specifically in The Inmates are Running the Asylum (coincidentally also using an ATM example) that a source of poor UX is that computers do not interact i.e. "converse" like real people. e.g.

  1. You: "Withdraw $200"
  2. Computer: "Insufficient funds"
  3. You: "Check balance"
  4. Computer: "$187.34"

with a human

  1. You: "Withdraw $200"
  2. Person: "You have insufficient funds, maximum you can withdraw is $187.3"

Now it is quite easy for someone to overshoot and think that a "conversational UI" goal implies "anthropomorphism" which does not necessarily hold.

In the example you give it seems an interesting case where computers don't do Conversation Alignment.

Now we must challenge ourselves to design systems that accept and support users’ conversations. Machines cannot yet negotiate alignment, but they can and should help their users carry on conversations

Something humanistic and a natural conversation UX to one Persona may be creepy for another. UI isn't adapting to contextual clues.

  • So basically you're saying that machines referring to themselves as "me" and "I" is the result of an overenthusiastic attempt (on the part of the designer) to increase interactivity?
    – dennislees
    Commented Nov 18, 2014 at 15:46
  • 2
    Essentially yes. Specifically in that case, I'd hypothesise the goal may be to make the flow of the interaction match that of a normal pleasant interaction with a teller at a counter.
    – Jason A.
    Commented Nov 18, 2014 at 17:18

I would like to give you one explanation that comes from my field of activity. It has to do with the design of human-computer-interaction or better: human-automation-interaction. Very often tasks are too complicated to be fully automated. At certain points of the solution process the user has to be incorporated.

Researchers found that a "mixed initiative interaction" is most suitable for this type of tasks. In mixed initiative interaction the system acts an an "agent", that is as an equal partner to discuss with during the solution process. This strategy aims to support and encourage the human decision making and problem solving skills.

With the concept of mixed initiative interaction in mind, the human-like communication style in user interfaces can be explained. It is often the most intuitive way to simulate the artificial agent working on a task together with the user. Below you find an article on this topic. It contains an example (the software "Lookout"), where human and computer negotiate about appointments. The user states his preferences, the computer checks the calender and confirms or makes alternative suggestions. The communication style of human agents is used, for example the artificial agent says: "You will be busy at that date, should I suggest another date?" This contributes to a natural flow of information exchange between two agents that aims to find a solution that is acceptable for both parties.

Summarizing, you can simulate a personality if human computer collaboration is essential for accomplishing a task and if a human-like dialog supports the human decision making process. Depending on the particular application there can be different ways to design mixed initiative interaction.

The cash machine example can be related to mixed initiative design as follows:

  1. At the point where the message appears the successful completion of the task not only depends on the machine, but also on the user, because the transaction cannot be completed if the user interrupts the machine or if he goes away.
  2. Therefore the user has to be prompted to wait and in order to make this comprehensible a reason has to be given ("the server has to be contacted").
  3. The I-perspective further increases the trust in the system and the confidence in a successful completion: It leads to the mental model "Ok, there is an agent doing its best to provide the desired service to me, I do what he asks from me". It also reduces the technical nature of the message. If the message was for example "Please wait, the system has to contact the server", the technical nature of the message could intimidate the user, he could even take it for a problem.

Reference: Hearst, M. A. (1999), 'Trends & Controversies: Mixed-initiative interaction.', IEEE Intelligent Systems 14 (5) , 14-23 .

  • Thanks Anna, I hope you don't mind my edits. The content was great, just really long. It took me three tries to get all the way through it : ) +1
    – dennislees
    Commented Nov 19, 2014 at 3:58
  • I suggest a compromise (see above) :-) The concept of mixed initiative interaction is my key to explain the human-like style of dialogs in user interfaces. Commented Nov 19, 2014 at 9:29
  • If you insist, but I think this is a better answer than the others and suggest one of the reasons it has fewer votes is that it is much less succinct. Less to-the-point. It's not very scannable, and the key concept is hidden in a summary paragraph halfway through.
    – dennislees
    Commented Nov 19, 2014 at 14:23

Why we chose to do something similar was the result of internal studies. Something we try to keep in mind is that we (designers|programmers|analysts) are not representative of our audience. Not to be snobbish, we just remember that half of our users are "below average" and we have to keep systems accessible to all. We aim for about a 5th grade reading level, which keeps it simpler for everyone, including those for whom English is a second language.

  • 1
    I didn't ask 'Why did X do Y?", I asked "Why is Y done?" - there's a big difference. A potential answer to the second question is "because research has shown that it positively affects... etc."
    – dennislees
    Commented Nov 18, 2014 at 3:26
  • Sorry about that, I removed the line. And we do internal research on these, aimed at our target audience; but nothing I could share. Commented Nov 19, 2014 at 3:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.