I am writing a novel with an ensemble cast. Each chapter is told from a different person's point of view. Some of the chapters are about a non-sentient machine that communicates with the human characters. It's not a sentient android like R2D2 from Star Wars or the killer machine from the Terminator movies or Data from Star Trek. It's no more self-aware than a very sophisticated machine, but many of its actions and communications drive the plot and decisions of the other characters.

This character will never become sentient during the story. It will never have feelings the way your laptop never has feelings.

How would I write from its point of view?

Edit to Add:

  • This character only seems intelligent and sentient.
  • It does not comprehend human motives.
  • It considers human requests baffling even when it understands the requests.
  • It seems extremely intelligent in that it has access to extraordinary amounts of knowledge and can even put that knowledge together in unpredictable ways.
  • 1
    If it's not actually sentient, how can it consider human requests baffling?
    – user18397
    Apr 18, 2016 at 9:44
  • @Thomo "Baffling" is not defined the same way for a non-sentient computer as it is for a human. If you type a formula with an error into Excel, Excel can be said to be "baffled" by it, and it will spit back a dialog box with an error message. Essentially, you have given Excel an instruction it can't follow. So it "understands" the request (instruction), but it can't execute it. Thus, non-sentient computer bafflement. Apr 18, 2016 at 13:25
  • 2
    Interesting, I have never heard baffling defined as that. What you described has always just been called an error, usually the result of an incorrect formula that excel can't execute. It's a yes/no thing. The operator might be right but the syntax wrong, it's not baffled it just can't compute it. The word consider lends itself to describing sentience, somewhat like HEX in Discworld, something which the OP seems not to want
    – user18397
    Apr 18, 2016 at 21:17
  • @Thomo If you enter incorrect formulae into an Excel spreadsheet, I expect the Excel program to say is an error. Excel won't be baffled. Imagine a program like Siri, but with access to enormous amounts of knowledge all over the world. Siri is not sentient, but it seems knowledgeable. It learns that some requests keep following certain patterns, leading the machine to put the knowledge together in unusual ways and then to gain insights. And the machine could learn enough to ask, "How did humans know to lead me down this breadcrumb of bits of knowledge to this insight?"
    – RichS
    Apr 19, 2016 at 6:35
  • 2
    @RichS Excel was the example given by Lauren. Siri is basically the same though, you quantify your argument and it responds within a predetermined set of parameters. The OP doesn't want a sentient program or an AI, I merely queried that if they want to avoid sentience, then it's impossible for the program to 'consider' behaviour
    – user18397
    Apr 19, 2016 at 23:03

4 Answers 4


A few possibile viewpoints:

  • An omniscient narrator who describes what the machine does and says.
  • One of the nearby sentient beings, when any are available to observe the machine's important actions or communications.
  • Reports from someone who pieces together the machine's communications and actions from available evidence after the fact.

Write it as data inputs and responses.

INPUT: USER 1 enters room
OUTPUT_$content: {greeting}; {Salutation: 'Good'} {TOD: 1415, 'afteroon'};
INPUT: USER 1 response {"Good afternoon yourself. Did you finish compiling that report?"}
SEARCH_DB6b.46: report {SMITH, CHARLES: activities prior 72 hours};
ET COMPLETION: 4.7 hours
OUTPUT_$content: {apology}; {report SMITH, CHARLES} {STATUS};

and so on.

  • 3
    I guess this could work for a certain reader, but I'd skip ahead when I came to this. Apr 19, 2016 at 13:17
  • @KenMohnkern Yes, it can be an acquired taste, and I myself would make those short chapters. But I think it would be the most accurate representation of machine thought from machine perspective. Apr 19, 2016 at 14:23
  • A very acquired taste. Like Ken, I'd skip it as it becomes too difficult to read. Even a short chapter would be too long, in my opinion, and you'd have to be careful if it includes important development. As the bookend to a chapter (i.e. start and finish with a passage), I can see it possibly working, with the rest of the chapter being from the POV of an observer noting how it reacts to different situations maybe?
    – user18397
    Apr 19, 2016 at 23:02
  • As a former software architect, I would definitely skip a chapter that read like that.
    – user2686
    Apr 22, 2016 at 18:59
  • 1
    +1 because it's a very cool idea... but it would definitely have to be a single chapter and extra short. Or there could be a moment of this and then the rest gets translated into more everyday English. Feb 17, 2017 at 16:45

You don't need to overthink this. Readers will accept whatever reality you present to them so long as it is consistent.

Just create a set of rules for the robot's AI then write the character as you would for a human. For example:

  • It can only use 100 basic words and key phrases.
  • It will only process the world as raw data. It doesn't see colors or humans; it just sees shapes and movement.
  • It doesn't ever make a personal judgement about the world. It only comes to logical conclusions based on its programmed parameters.

The fact you're expressing its thoughts in English and not binary code doesn't really matter. Readers won't question its sentience as long as you follow the rules of its AI.

What you need to keep in mind is that a machine really doesn't have sentience. So you can't truly express its thoughts in words without giving it a veneer of human consciousness. But just as we see things on a computer screen translated from code, so too will readers assume the English words they read are translations of code.


What is your story purpose for giving the machine POV? Why does the reader need to get inside the machine's "head?"

If you want to show the machine's limitations, it can be done with a POV human struggling to get the machine to understand.

Now if you intend for the machine to make an important mistake at some point, then I can see using the machine's POV.

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.