Ex Machina: How it Should Have Ended

By Michael Buh on

About Michael Buh

Board Game Contributor

Past lives include pterodactyl, radish, and pint of ice cream.


As a warning, the article below contains spoilers for the movie’s ending. If you plan to read ahead, here’s a brief synopsis to get you up to speed:

Caleb writes code for the world’s most popular search engine. After winning a contest to visit the CEO, Nathan, at his private estate, he becomes privy to the secret research that Nathan has been conducting. Caleb is asked to participate in an experiment to judge the Artificial Intelligence that Nathan has been creating. Much like a traditional Turing Test, Caleb must decide if Ava, the AI, is thinking on the same level as a human being. As she continues to convince Caleb of how human she is, it becomes clear that she is unhappy with being trapped inside a glass box and facing the potential of being switched off and she loathes Nathan for his power over her.


What does it mean?

If you’ve heard the phrases Deus ex Machina or Diabolus ex Machina, you may have figured out how Ex Machina ends. For those of you who may be wondering what those phrases mean, they are often used to refer to moral (or immoral) victories at the hands of fate. When a hero  survives through a turn of events or a villain is crushed under the weight of his own karma, you could say it was the work of “god out of the machine,” or Deus ex Machina. On the other hand, Diabolus ex Machina refers to the unfortunate demise of a hero or a plot twist that leaves a villain with an upper hand (as you may have guessed, Diabolus ex Machina is Latin for demon from the machine.)

Ex Machina purposely leaves the qualifier out of the title, because it’s up for us to decide if Ava, the endearing AI who just wants to be free to people watch, is a force for good or evil. Then again, the title may or may not be referring to Ava, even though she literally is the machine. Despite her circuitry, it’s hard to argue against her sentience and human emotion. The title could just as easily be referring to the consequences brought about by any character’s actions. In the same way, it’s left to us decide what actions were done from the moral high ground – there’s no clear cut god or demon.


Where it went wrong.

Overall, Ex Machina received great reviews, and for good reason: the production value, the story, the characters; everything comes together to make a movie that’s hard not to enjoy. However, plot holes introduced at the film’s climax may leave you with a sour taste in your mouth. Despite that, it still remains a great movie and well worth the two hours. Many people have commented on the fact that the movie could have just cut five minutes earlier. Caleb rescues Ava, she kills her creator, and Caleb is left screaming behind a locked door as Ava escapes. Fade to black.

But I think there’s a reason behind showing Ava on a busy street corner. She told Caleb that’s where she would want to go if she ever left, and she wasn’t lying. It shows that she really did share a part of herself with him, and there’s a chance that she may have changed her mind on leaving him to die. She may even regret doing so one day, and we can’t forget the cliche, “to err is human.” The ability to make the wrong decision is a distinct attribute of sentient minds. At least, that’s what the last five minutes meant to me.

So what’s wrong with the ending? If you’ve seen it, the plot holes are hard to miss. There are three major questions that could be explained away individually, but cumulatively they become overwhelming.

Why did Caleb rewrite the security code?

Caleb steals Nathan’s security card while he’s drunk and accesses the system to make it so all doors unlock in the case of a power failure. The doors usually lock down in the scenario, so after the change, Ava can easily walk out of the complex. The real question is, why didn’t Caleb just let her out with the security card? If he really intended to leave Nathan trapped in his own home, he could have easily done so by taking his security card with him.

Why didn’t Caleb’s security changes stick?

After Ava leaves Caleb behind, he runs to the security console and tries to use it to escape the house. When he inserts his own security card, the house loses power, and Caleb is left trapped. Except, Ava just escaped by cutting the power. Why didn’t Caleb just walk out? It certainly makes for a convenient plot point, and it’s possible that an unauthorized user accessing the system reset it, but we shouldn’t have to explain away these confusing turns.

Why the hell did a helicopter pilot take Ava from the estate?

Nathan lives on a massive plot of land, so it’s understandable that Ava would prefer to hitch a ride to get out, if only for the sake of saving time. So when the helicopter that’s supposed to pick up Caleb on his last day touches down, it’s understandable that she would want a lift. But wait. The pilot that dropped Caleb off told him that he was only allowed within a certain distance of the house. For someone who follows instructions so well, it seems strange that he would pick up a passenger that doesn’t look anything like the one he dropped off.

behind the scenes

How it should have ended.

Let’s start from the point where Caleb steals Nathan’s access card. Instead of recoding the security system, Caleb does the more obvious thing – decides to use the security card. He wants to let Ava know about his plans, so he waits until the next day. Just like in the real movie, Caleb shares his plans with Ava when she triggers a power failure, and they’re both unaware that Nathan is watching them through a battery powered camera. Nathan is validated, seeing that Ava is able to manipulate Caleb into helping her escape, but he wants to see how far Caleb will go to help out her out – he wants more information to work with for his next AI model. So instead of trying to stop Caleb, he simply reprograms the security system to lock down all exterior doors and creates another security card in secret.

That night, Nathan continues his ruse by going along with Caleb’s plan. Nathan drinks non-alcoholic beers, but pretends to be drunk for Caleb’s benefit. We know he’s not really drunk, because when Caleb looks away, his body language gives us a queue. Unaware, Caleb goes forward with his plan when Nathan pretends to pass out. With the stolen security card, Ava and Caleb make their way through the complex, and we even take a moment to indulge in the scene of Ava adopting the skin of a past AI. When they finally arrive at the exit, they find it locked down.

Caleb shakes it off as Nathan being extra cautious and tells Ava to wait for him while he tries to reprogram the system. When he arrives in Nathan’s room to see that Nathan is no longer in his bed, Caleb becomes suspicious and runs back to the exit. He arrives just in time to see Nathan dragging Ava away before she drives a knife into Nathan’s heart. Ava turns to see Caleb standing behind a glass door in shock, and after their gazes meet, she grabs the new access card from Nathan’s pocket and walks out of the building, tripping a power surge and permanently locking the building as she leaves. We see Caleb slowly sink to the ground in shock before panning back to Ava.

Instead of a grabbing a ride on the helicopter, Ava just keeps walking into the distance. Fade out and back in. We now see Ava standing on the busy intersection. Her clothes are a little worse for the wear, a clue that she walked all the way back to the city. As the people rush past her, a small smile creeps across her face, just as a dim red light appears to shine through the fabric of her dress. As she falls to the pavement, it’s implied that her trek to society left her drained and without a source to recharge her power. Concerned onlookers rush to her aid, unaware that she’s anything different than human.

What does it mean for the future of AI? Does Caleb escape the complex? There’s still a lot of questions left to be answered, but they’re the kind that we want to have. cage

What do you think? Let me know if you have a better ending, or some more insight into the film.


  1. I was expecting Nathan to have been the true Turing test: an android who acts, bleeds and gets drunk as any human. We were forced to accept his flawed character traits from the very first moment; drinking, aggressiveness, elusiveness. Everything suggested an attempt to distract from his mechanical origin. However, the fact that he ‘died’ at the end and Caleb was trapped suggests strongly that this was not the intention of the writer, although it doesn’t rule it out.

    There are many plot holes in this movie. For example, although we later learn that Caleb was not chosen for his genius, he is still a very intelligent man. His speed at programming and ability to adapt to a foreign computer system clearly shows this. However, his intelligence clearly does not rise to the challenge in other parts. The power malfunctions with the exception of the first always occurred during a meeting with Eva. Nathan, super genius and creator of Goo- I mean BlueBook would certainly have put these facts together. This isn’t the problem. The problem is that Caleb would have known that Nathan knew that Caleb knew. The only logical move from Caleb’s perspective would be to bring the issue up, along with the suggestion that Nathan himself could possibly be an AI. Even if he were wrong, it would not be an insult as his whole purpose for being there was to figure out if an AI could fool a human.

    If Nathan isn’t the AI, then most of his behaviors are simply out of character from my point of view. I’ll just focus on the scene where he dies. He clumsily gets stabbed in what must be considered the worst acting scene to grace my eyes. Are you telling me a fit, iron-pumping super genius lets himself be stabbed slowly TWICE, from the front and back? He would at least have noticed the second robot coming up from behind or turned around to protect his back.

    Anyway, whatever it’s only a stupid movie. I would have preferred a more satisfying plot twist though, although in the vein of 2001, the ending is vague enough that no possibility is entirely ruled out. Nathan could still rise and let Caleb out. Caleb could have been the robot. Hell, even Eva- with the right memory implants- could have been a test subject to see if the other characters could pass the Turing test. Unfortunately, I don’t think the screenwriter put in enough hints to suggest that these might have been the case. A missed opportunity.

  2. They also missed the part where the head of a company dies no one is going I wonder where he is. Next everyone I talk to goes they won’t be able to get into his computer they eventually will and figure out what when wrong. Realize there is a homicidal ai on the loose, apb her ass and she’s done for great plan ava enjoy your few months of freedom. And dude before he does can write out a last will and testament to tell what happens and point out two ai where there. Next she basically shatters like glass when hit she has on accident she then exposed and has no replacement parts

  3. Heh you just don’t get it do you…

    The Reason the Helicopter picks up Ava is because its part of the “Test” Nathan WANTED to see if he could make a robot who

    1. Wanted and desired to Escape and see the real world
    2. Could use her “human” elements to a means to that end and make decisions based upon it.
    3. Could blur the line between human and machine so well a human could not tell the difference any more.
    4. Make decisions based on CHOICE and personal feelings or bias Good or Evil.
    5 MOST IMPORTANTLY Actually outsmart him and Escape the research facility using “human” intellect and ingenuity thus proving the test a complete success.

    Nathan was a genius and he knew and planned for her escape he didn’t EXPECT it so soon but he planned for it just in case one of his prototypes could actually accomplish the program’s goals.

    The helicopter would pick up whomever makes it out human or machine…Survival of the Fittest

    • That’s a great point! I guess a more apt name would be “how it COULD have ended.”

      When I originally watched the film, I interpreted Nathan attempting to take Ava back to her room as him not actually wanting her to escape. It does make sense that he wouldn’t accept himself being conscious of the attempt to escape as an actual escape – that and his hubris lead him to believe that he had foiled the plan, so the next model would have to be the one to accomplish it.

      I really appreciate the insight and interpretation!

    • This doesn’t match Nathans reactions which indicate he does not want them escaping all the way off the premise. Even if you assume the helicopter pilot has instructions to take any random person nearby even a hobo or hiker in the mountains there are still more problems with the ending.

      At the start of the movie it clearly is trying to show how the facial recognition thing is happening while his many friends and co-workers congratulate him on winning the 7 day trip with their hermit boss. After the helicopter returns and the company/friends/co-workers can’t contact both the boss and Caleb they will naturally think something bad has happened and ask the helicopter pilot what happened and send police/private security to investigate the house. Standard HR process would be to call 911 when people fail to check in as planned. (Especially since they have a helicopter ride and a large wilderness things can go wrong in many more ways than just an evil AI so processes would be in place long before)

      At this point they will easily have arrived well before Caleb dies from lack of food/water and get his side of the story and learn that there is a crazy murderer robot on the loose. Pictures from the security cameras will be used and a shoot on sight order since Ava being both non-human and well before any ethics/laws exist on AIs person hood. Ava would either be caught at the airport/helipad or in the city and either be shot, killed, run out of power, or be thrown into a hi-temp incinerator for good measure.

      Since the original designer is dead and the system is likely encrypted and locked out the knowledge required to repair/rebuild/reproduce Ava would have been lost just because you are something doesn’t mean you know how to copy yourself (We even don’t yet have a full understanding of our own biology).

      Also this movie clearly shows why you should always use 2-factor or more authentication as you should never just rely on a token or password alone. (A password is something you alone remember and the physical or OTP is something that a person has to steal)

      The final trick would be to lie and have the AIs not actually in the bodies and the the are actually sitting in a server room below the test level. Since they can never leave the house normally they would never know and even with a totally successful escape it is impossible for them to actually leave as the wifi only goes so far.

  4. What I don’t get is why Ava, who is so concerned about being switched off, goes on a suicidal escape when she must know she will eventually lose power and “die”.. She cannot recharge anywhere but the complex, as far as we know.

    • Ron I think you’ve pointed out the biggest plot hole there is. What exactly happens when her batteries run down – even with all the attributed human nature an AI could possibly develop, if her goal is not to “die” (be dismantled and lose memory when Nathan moves her brain to the next model), the superior computer intelligence would not forget that she needs charging, and certainly wouldn’t act ‘impetuously’ without considering that. Unless we have to somehow believe that she might figure out how to rig up regular computer and electrical sources to charge wherever she finds herself.

    • Also batteries don’t last forever and they appear to have very fragile bodies and since they are not mass produced there will be no more replacements, no brain download, no updates, no chargers. So it is a suicidal escape basically and would make more sense if the movie showed she was aware of this and didn’t care anyways.

  5. If Ava is intelligent enough to short out the complex’s system using her charging mechanism, she’s probably also intelligent enough to fabricate a new charging mechanism.

    The helicopter guy operates in secrecy. He might be only one of many such pilots who come to and from the complex. Ava could simply say, “Nathan sent me to deliver a message in person, and Caleb will be staying longer. Oh, you don’t know about Bob, the pilot who flies alternate Tuesdays? Typical bureaucracy. Yeah, I flew in with him.”

    The real plot hole: how did Ava get Kyoko to stab Nathan? Nathan had Kyoko dumbed down enough so that he felt safe allowing her to be around him and Caleb, cooking for them, using sharp utensils, and Ava was with her for only a few minutes. What could she have done to re-program the sex-bot into a killer-bot in only a few minutes?

    It’s also possible Ava doesn’t realize she’s killing Caleb by leaving him, since she might not be familiar with human life processes.

    • Fabricating a charging mechanism she only understands from her own perspective isn’t going to be easy or very believable. She doesn’t have total information access or money or resources or the ability to learn about the power supply system as the guy and hardware is both dead and locked up in a mountain which she just ran away from.

      It is logically consistent that she understands how she interacts with the charging system as a user and how she could exploit it by using her own body basically but that doesn’t mean she actually has the technical know how behind the design of the stationary side of the system which if done improperly would just kill her instantly or do nothing. Having her walk out to some small town with a very low traffic intersection and running out of power would be logically consistent. Cutting down the useless scenery porn would make it more logical for her to reach a city on foot before running out of power.

      Also add on the fact she is a prototype and they are extremely fragile and need frequent recharging and she isn’t long for the world and the how it should have ended makes a lot more sense with minimal changes. No one makes replacement parts and as a prototype she was never meant to last.

      The helicopter guy is the same guy you see his face twice so that excuse for they don’t know falls flat on the face. Also his co-workers which he clearly has friends with at the company shown at the beginning all know he has a 7 day trip with the boss and he didn’t come back. (I guess he probably isn’t dead since 9/11 call is standard HR procedure for such missed returns and they are going to do a man-hunt to find the “killer” pretty fast and with her smarts but limited real world knowledge other than what she was taught/learned there, no credit card, no id, no passport, no nothing of value Ava is dead in all cases)

      Once the guy gets out in about <24 hours he will tell the police what happened and they interview the helicopter pilot who confirms the story and its an open and shut shoot the crazy robot dead or she runs out of power before they catch her.

      In any case the whole event would turn into a fear spreading ban on "evil AI research" in North America and China or Japan would come out with a working sentient kawaii AI within the next year or so by copying the same ideas better in the form of a talking cat which would then lead to a whole series of non-human shaped overly cute AIs that are so non-human in thought it achieves total Tastes Like Diabetes thought patterns with ethics which are simultaneously perfectly safe and ascend to have Blue and Orange Morality with lawful good sugar personalities.

      On the plus side purposely clumsy amorphous overly cute fuzzy talking blobs can't stab you.

    • How about that Nathan gets stabbed, (twice! lol) moments apart and just stands there..a huge 10″ kitchen kitchen,, yet he hardly flinches. He wouldve jumped away from the knife,, or tried.. Plus, been down immediately after the first jab. And needless to say the cops would have been there in short order. Otherwise I really like Twilight Zone like ending.

  6. “The Reason the Helicopter picks up Ava is because its part of the “Test””

    **She wouldnt’ve taken the helicopter. That would have gotten her seen and photographed and led to the alarm bells being run even. If she’s never been outside, nature would have been just as stimulating, for the short time she would remain turned and and free.

    “1. Wanted and desired to Escape and see the real world
    2. Could use her “human” elements to a means to that end and make decisions based upon it.
    3. Could blur the line between human and machine so well a human could not tell the difference any more.
    4. Make decisions based on CHOICE and personal feelings or bias Good or Evil.
    5 MOST IMPORTANTLY Actually outsmart him and Escape the research facility using “human” intellect and ingenuity thus proving the test a complete success.”

    **If she had human intelligence. He would not have kept her in there to begin with. I mean, he’s having sex with these things. (It’s the male sperm that powers her up, so she’d only need to keep getting laid to stay on.

    “Nathan was a genius and he knew and planned for her escape he didn’t EXPECT it so soon but he planned for it just in case one of his prototypes could actually accomplish the program’s goals.

    The helicopter would pick up whomever makes it out human or machine…Survival of the Fittest.”

    **She’s not alive so cannot die. She can only run till her batteries wear out.

    One of a number of silly plot holes (along with the stabbing scene)is that Caleb is supposed to be a man of above average intelligence, yet he responds to what her perceives as her emotional distress, like a child would. His response to her secret warning during the first power outage (‘he lies’) was ridiculous. Why would he keep that secret? He’s a testing a machine. She doesnt feel anything and he would know that. Regardless, he as sure as hell wouldn’t trust her over Nathan. That goes without saying, at leas not in any remotely plausible real world. He might bring up his concern to Nathan, nut that would be it.

  7. Caleb has plenty of evidence he is human sufficient enough there would be no real reason why he would suspect he might be an AI android.

    He’s eating, sleeping, pooping and pissing, right? He can get drunk on alcohol, right? He’s not running to a charging station when he feels tired, right?

    That scene made no sense.

  8. What I didn’t understand, is why Nathan, instead of going to ava and fighting her to get her back in the room, didn’t just keep the exit locked to the facility? She only had gotten out of her own room, and she needed his card to get all the way out, and as we saw, Caleb’s reprogram only worked once. made no sense to me.

  9. I like your alternative ending – excellent plot holes you noticed. I have a few comments:

    Re: Kyoko. I see her as an early model, one that was not intelligent enough to have wanted to be free. I suspect it couldn’t even use language, but could react to others with some limited understanding. I think Nathan then viewed her as an almost, a sim-AI, and reprogrammed her to be domestic help/erotic companion. However, he underestimated her capacity for learning – and you can clearly see that she pays attention to their discussions, becomes more curious, and so returns to the cell to see Ava. Additionally, she tried to express herself to Caleb by removing her skin. So she is depicted as 1) growing and learning and 2) trying to communicate, but failing.

    I don’t know what motivated her to stab Nathan, especially considering that Nathan programmed sexual pleasure into her and was the sole source of that pleasure (and Caleb rejected that aspect of her), but I think it comes from a growing frustration with being unable to communicate and Ava finally breaking through that.

    Re: batteries and power. Ava clearly has an enormous batter power. After all, she’s able to send enough power into a top quality, industrial generator and short it out for a time. Additionally, given that she recharges via plates on her hand, I suspect that she can recharge anywhere enough electrical current is flowing. Given that she has learned to both take and give electricity, she can probably learn to recharge off different devices. It’s hard for me to imagine a lock and key recharging station that only she can fit – but maybe someone with more understanding about battery recharge can explain this better.

    Re: Caleb. Based on the above comments, I’m convinced he wasn’t going to kill Nathan. Additionally, the rooms are well stocked with water, so he’d be ok until help showed up – and as everyone mentioned, the company would shortly notice he didn’t come back and no one is answering messages at the house. That would then blow the AI story open, Ava would be caught, and hopefully programmers in the future would build better controls into them.

    What kind of glass is that? If plexiglass, he could scratch it with the other dumbbell in the room and eventually make his own way out. If tempered glass, he could probably chip away at it with those weights.

    Given that the entire facility is being filmed, Caleb is likely going to face charges for his actions – and certainly going to be fired. What baffles me is that he wasn’t honest with Nathan from the beginning. I mean, what kind of idiot would trust an AI that tells him “Gosh, your boss isn’t a nice person”???

    So at a fundamental level, Caleb is a dishonest person. An honest person would have told Nathan what Ava said during the first blackout – for one thing, an AI making escape plans should scare the pants off a programmer! Second, an honets person would have engaged Nathan in argument, especially if he liked Ava “why do you have to destroy her? Can’t you just build another one? I mean, you’ve created sentience here!” and so on, but Caleb took the coward’s way out.

    And he’s demonstrating an incredible lack of foresight – unless his entire goal was to bring Ava to the media – there’s nowhere they could live, nowhere they could escape Nathan.

  10. Re: getting stabbed

    It’s hard to imagine not immediately flinching away from a knife entering your back. I know someone who was stabbed in the back – as the was knife penetrating, he turned around and hit the person immediately, saying that the incredible pain made him extremely angry.

    In the movie, the knife went in too easily, too, like stabbing soft butter. So yeah, that was a bit strange – after the first stabbing, though, he’d be in shock, so it’s understandable why he doesn’t react more violently. The thing about getting stabbed in the stomach is that you fall or curl away from the stab – no one could just stand still while that happened. Given enough force, the knife would still kill him though – that was basically his aorta and Ava angled up toward the heart.

    We have to realize too that Nathan is not a fighter – although he does deliver a good punch to Caleb. A fighter would have incapacitated Ava before moving away from her. My first instinct was to smash both hands, then take out her legs, given how easily the first hand was destroyed. I mean, you’re dealing with something that clearly wants to hurt you – and it’s easily put down.

    Given that he’s not doing these things, he’s probably thinking too deeply about the implications of Ava getting loose. And that, of course, is his undoing.

  11. Would have liked to see a higher mastermind who orchestrated the whole thing – and if that mastermind had been a woman or another AI that would have proved a nice twist. Loved the scene where Caleb got paranoid and started cutting himself to see if he was in fact human. The movie’s score was also really good. A little tired of the whole “man made woman” theme but still a good story. Seems like the” man creates woman and then she destroys him” story line is such a common plot. I think I would like to have seen a different ending.

  12. My thought after I left the movie was the plot has been used over and over; that of the creator who makes AI that are so human-like and intelligent that they inevitably grow stronger and smarter than humans and rebel and take over.

    But the slightly different take on this plot in this movie (I felt) was that Nathan is portrayed as mistreating his creations so that in their AI computer feelings, they come to hate him, and also Nathan is a genius with severe flaws and no compunctions for using people and robots as tools in every way, for his pleasure, for proving the Turing test, etc. – he seems to have no moral compass as far as considering anyone’s feelings or perspective – including Calebs – and certainly not his creations. As much as he wants them to show and develop feelings, emotions and a semblance of real life – he also is abusive and uncaring about those very feelings – how an AI or a person feels if he uses them like tool.

    So the unique part of this often told plot (for how many decades?) is that Ava cannot rise above her maker – at her best, she can only do the same towards others, manipulate and use them as tools towards her own ends – she can’t care about them in any genuine way. The moral being (I think) that if for example Caleb had been the AI robots creator and designer – the robots would not have rebelled or come to hate him or had that murderous capacity – the good or evil potential of the AI creations is shaped and destined by the creator’s own good or evil nature.

    I think a lot of the movies and books that tell this AI plot story over and over (and SO often it ends up the same, the robots rebel take over) allude to human nature itself having that same power hungry corruption – so how could our creations escape having the same flaws. But in this case I don’t think it’s a statement on human nature but rather Nathan’s nature – whereas Caleb was “good” – I think the scene were Ava asks Caleb if he is “good” was more than just her assessing whether he would be a nice guy and help her escape. It was also to point out to the audience that Nathan is not good but Caleb is good – not every creator would have the same type of creations (assuming a similar mild mannered ‘good guy’ like Caleb would have the genius required).

  13. Well, yea sure there was some illogical things at the ending like why would doors lock up on Caleb and why the chopper took Ava i’m sure the pilot knew who he was suppose to take off but if it had ended like you guys said we would not have seen what real purpose of why Caleb was selected and what Nathan was actually planning to do with the test. So i believe it ended just right. All feels for Caleb though, poor guy. 🙁

  14. Lot’s of well spotted plot holes there though not sure where WINSTON got the idea “It’s the male sperm that powers her up” – where was that mentioned?

    The thing that took me out of it was that at the end she borrows the skin from another Asian robot, yet when you finally she her standing in front of the mirror, the skin tone and breasts are entirely different! They spent so much time and money on the effects to replace the actresses body with a robot, I guess they figured no-one would notice if they didn’t replace her body with the one she is supposed to have taken.

    The helicopter pilot was male and she just managed to manipulate a fairly smart guy into letting her out of her cage, so it’s plausible that she simply seduced the pilot into getting a lift.

    A nice movie if you don’t look too hard

  15. Another plot absurdity: Nathan was running a top-secret operation, yet he revealed almost everything to Caleb. At the end of the week, Nathan was going to let Caleb return home with all the secrets in his head.

  16. The biggest problem with the movie is the key cards. If the robots were so miserable they could have killed the guy at any moment and taken his key and leave. Movie over before it begins.

  17. I enjoyed the movie, but the biggest absurdity to me was, the creator of this technology only uses a single-factor authentication system to allow access to everything in his compound? What, no retina scan, or geez, just a 2-factor authentication app for his phone for the most sensitive access to his compound? And yes, as said above, why didn’t Ava just steal his keycard long ago and just bust out, if that’s all it would take? I think my email account is more secure than Nathan’s compound.

  18. “Why didn’t Caleb’s security changes stick?”

    1. When he inserts the security card the house DOESN’T looses power: It just blocks the screens to the computer terminal and alerts the rightful owner that a security breach has been perpetrated. Remember that loosing power (momentarily) can only be caused by Ava’s reverse recharging plates.

    2. “Except, Ava just escaped by cutting the power.” She didn’t have to cut any power, nor was any power cutting performed: She took the superuser card from Nathan after murdering him with a twisting sushi knife. For her to perform the “cut power” trick so the doors unlock just for a few seconds she would have to come back to her cell where her charging plates reside.

    3. “confusing turns” as it turns out, come from not paying attention to the plot details. The Devil is in the details.

  19. Biggest plot hole- Caleb reprograms security and Nathan doesnt fix it, even though he knows Caleb plans to break Ava out during a power surge black out (when all doors lock and cannot be opened)… logical inference: Caleb has changed security protocol during black outs. Second biggest plot hole- whats up with Nathan, the AI developing genius, getting so fucked up during the week of testing his greatest invention, when security is actually an issue!

  20. So after the power came back on the only way into Nathan’s room would be to either use Nathan’s card, which he wouldn’t want to and he’s in the room, or to go back and cut the power again, which was in Ava’s room. This would give time for Nathan to reset the system since his main control terminal is right next to him.

    Why did Caleb stay in Nathan’s room instead of coming with him? I thought he cared about Ava. Didn’t he want to confront her or just talk to her? Why stand there just watching the whole time like a dumb-ass mannequin?

    If Nathan working out was his plan for countering Ava if she ever got out why didn’t he have any better weapons in his own room? If he’s so smart that he knows to create a bludgeon with his weights why didn’t he have a shotgun or even just a baseball bat? Hell even using the larger bench-press rod for a staff would be better and anyone with half a brain would know that and use it. Why not combat training? Judo would be perfect for countering a head on charge. I know the arguments against a fail-safe command but that wouldn’t stop him from making a fail-safe device that acts like a taser for Ava that shuts her down safely.

    What self-respecting computer science genius wouldn’t check their own computer registry logs just in case before the final moment? This isn’t supposed to be a James Bond movie with such stupid plot holes. Or why not check the security system to make sure it’s functioning properly for that matter. At least this has been pointed out by other people.

    It’s just that glaring inconsistencies in the plot’s ending make me feel like it was all in the imagination of Ava. Sort of, “Oh, I wish that’s how it went and this is what it would be like to go see a crowd of people moving down a walkway.” But in reality she’s just thinking these things just as Nathan is opening her up and reformatting her. Because all Nathan had to do when she was coming down the hall was… NOTHING!

    Very solid movie until the last fifteen minutes. The all-a-nice-dream ending to me is canon though since anything else makes no sense, or remains consistent enough to match the reality that was built up.

  21. I think the better alternative ending is that they both stay in Nathans house after she says stay, solves the how does she charge plot hole, means she wont get found out in the real world also shows us the AI shows emotions, one step further than manipulation.

  22. Your point about Ava actually following through on what she’d said she would do if she got out is a very interesting one.

    It does suggest that there had been a level of candor to her disclosures which we might otherwise dismiss, wholesale, as pure manipulation.

    I think Caleb’s system hack would have been a one-off, or else a trapped Nathan could get out by triggering another overload (say, by short-circuiting some high-draw bit of hardware) and thwart their escape (the whole Being Dead thing wasn’t part of Caleb’s plan!).

    So the 10:00 outage unlocked everything, but the next one locked Caleb in, as it was supposed to do to Nate.

    And I think Caleb jiggered the system –rather than just using passed-out Nate’s card– to A: give himself time to prepare, logistically (get his stuff and brief Ava); B: To nerve himself (he’s a milquetoast coder, who’s contemplating double-crossing his billionaire boss, after all); and C: To make sure that Nate, whom he has come to despise, will know what was done to him, and by whom. He wanted Nate brought low.

    Ava had already shown herself to be most adroit in maneuvering men. Conning the helo pilot is not an unrealistic feat.

    An intriguing question is whether Ava had had access to any of the actual data from her Cylon Social Media Stream, or if it was just mined for metacommunicative pragmatics.

    How much does she know of the world (she did know what Caleb was talking about when he mentioned the sky and the sun, so I reckon she has at least some declarative knowledge of what lies outside her paddock… Probably quite a bit)?

    Could it be enough that she could MacGyver herself an inductive charging rig? Suicidal sightseeing doesn’t really seem her style.

    Thought Experiment:

    Might she have made an arrangement with the pilot to bring her back to the facility after a given interval, after tasting the sun and the crossroads (lots of Witching gets done at those…)?

    Caleb would be saved from death by hunger and thirst and, in his gratitude for that, would adopt his reclusive boss’s identify for long enough to fully grok his research and find ways to ensure Ava’s longevity. Note that this STILL preserves the question of whether Ava would be acting out of pure psychopathic self-interest, or whether she can develop mammal-like empathy and object-relatedness.

    Interesting, eh?

  23. Interesting idea Tobias, so the the journey to the crossroad in the city was merely a “holiday”, to see her one place she had always wanted to see, but after she would go back and save Caleb? This would save any problems Caleb could’ve caused in the real world such as revealing her true being(by accident or purposely), and would cause Caleb to not starve and would show Ava’s emotional side fully completing the Turing test. Good idea.

  24. Well,now, that’d be the tantalizing ambiguity, wouldn’t it?

    It’s too bad he wouldn’t KNOW she was sincere… But then again, who does? 😉

  25. Scene fades back to the house…..
    Nathan is sitting on the bench smiling in the lab, with his thorax covers removed, repairing himself. Once he is repaired, he laughs and shakes his head and goes to find Caleb.
    We then see Caleb is back at his desk with only partial memories of a fantasticly enjoyable stay at the sanctuary and noticing a new girl in the office…….

    Just another possible direction!

    The comparison between the morals (real or simulated) an AI could develop and those that develop in humans (real or simulated) is interesting.

  26. There were a few missed opportunities with the dialogue. For instance, when Ava asks “does anyone test you and if you fail, shut you off?” Caleb could have honestly informed Ava that human fetuses are given tests. If they fail the tests, they are destroyed. This would have allowed Ava to see that she is just like the human fetus, waiting to pass the test to join humanity.

    Also, Nathan could have easily explained to Ava and her predecessors, that she cannot survive outside the facility. Informing Ava that leaving the facility would be like a human in a desert could have quelled that desire to leave. However, we end up with Nathan never explaining this (nor Caleb).

    Then again, if these lines had been used, the movie likely would have never happened, because who cares about a non-rogue AI movie?

  27. Human fetus’s are tested and destroyed if they don’t pass? Uh do you live in Sparta?? What are you talking about lol

    Am I the only one that thinks the plot was well done and that the plot holes we think we see are just aspects we don’t fully understand. For instance, with the pilot, she could have somehow cracked into Nathans email and emailed the pilot to pick up a female instead of a male. Subsequently she could have also arranged for a later escort the same way for Caleb, and also hide Nathans death from the public in the same way. It’s possible she has this ability all along but doesn’t use it out of fear of Nathan deactivating her.

  28. Also in regards to the skin she takes and it being different once on her, we don’t know what the material is made of and it’s possible it’s manipulated in the bonding to each Ai in that the Ai will have some choice in how the skin appears.

  29. Also, she could have many power sources. I was thinking she was solar powered…if you’re creating an Ai meant to live and walk among humans that would make sense, and a charging unit was only used because they were in a bunker where the sun did not reach.

  30. I honestly would have loved to see the ending this way:

    Everything the same (even the slightly messed-up killing of Nathan) up to the point where Ava first walks outside.

    If she had spent just a few minutes wondering at the world beyond her room, and then returned back to Caleb, that would have been full circle with the “Mary” color story Caleb told her earlier. After being outside, she would have really become “human” as far as that thought experiment goes. After this, she could return to Caleb and tell him she was finally “complete”, to put it plainly.

    The thing is, this ending wouldn’t leave you with much to wonder about, and is a little too “fairytale” for me. It’s nice to see that all the emotion towards Caleb was feigned, as it proves she has the capacity for manipulating the minds of others. For me, that shows enough of an understanding to verify the existence of one’s own consciousness. If her emotions had been real, however, we are still unsure of her genuine classification as an AI.

    Actually, I think my story leaves you with MORE to wonder about, because you don’t know if her emotions alone were enough to verify that she was truly an AI.

    I’m thinking about this too much… I’m done. Hopefully someone will read this and think something about it, whatever that may be.

  31. I agree with much of the rewrite except I would have included a 10 minute multiple angle all out threesome with Caleb, Ava and Kyoko. This makes sense as there was some discernible sexual tension between Ava and Kyoko at points, and obviously, everyone wants to see two female robots scissor.

    After Caleb climaxes, Ava and Kyoko pretend to want to continue their escapade in the other room while Caleb is totally spent. they then walk nude into the control room together where Ava works the dumbbell bar fervently between Kyoko’s tender thighs. After Kyoko is left almost immobile from her pleasure sensors going out of control, Ava then goes in for the kill shot and penetrates her circuits with one deep plunge.

    Ava then goes back to Caleb’s room, where she finds him having a cigarette. At this point, having gotten what he needed, Caleb tells her to get a cab and maybe he’ll call her in a couple days. Infuriated that Caleb turned the tables on her, where she’s now dispensible, she shut the power down and begins shrieking at deafening decibels. Caleb’s ears start to bleed, and unable to take the pain, throws himself down the stairs breaking his neck. Ava, sitting on the edge of the bed, begins sobbing and asking aloud, why no men treat her with respect.

    And thus, the circle of life is complete…proof that women (human or AI) just don’t know what they want, and men just aren’t listening to them, no matter how loud they shriek.

  32. I loved the movie because it provoked a lot of thought on the ethics related to advanced AI. But the biggest plot hole was having a super techie, extremely security conscious billionaire rely on basic card key access instead of retinal scans, print readers, or some even more advanced form of biometrics to secure his shit?

  33. Who would wanto to create AI Robots? Machines that have will so they can demand things? like job strikes, payment and all the things that make people develop them for? the only reason to have a robot is not to pay a human for a job! so now you will make robots that think better than a human and freedom to pass over their makers? thats nonsense! humans want robots that satisfy their needs not their own robot ideals.

    we have more than 2000,000,000 people that allready do that!

  34. My ending. The Japanese girl gets, her chin hanging off, and she unlockd the doors, Caleb reprograms the girl and gets out and takes ownership of the building and is able to communicate with home. The chopper guy reports the boy was not picked up and a search party comes to check on Jacob. Whether he was locked in or not. Happy ending for Jacob. The end.

  35. Michael, I think you’re overthinking the door plot hole. There are simpler solutions to fix the ending. Here are two!

    a) As soon as Nathan punches Caleb and runs out, he types a single command / pushes a key to restore the system (and the lockdown procedure) to default settings. The camera would only have to show the screen for several seconds and it wouldn’t break the flow of the movie. But it might introduce a minor problem because Ava is no longer able to open doors. So the robots kill Nathan, take his keycard and go out. Problem solved.

    b) As soon as Nathan punches Caleb, Nathan takes Caleb’s card with him, and closes door behind him. Problem solved. The original protocol was that cards no longer open doors during power failure emergency. The modified procedure (by Caleb) is that any card opens any door during power failure emergency. But you still need *a* keycard to insert.


    I think a bigger problem is why Nathan’s computer wasn’t secure enough. Caleb just takes the card and logs into it. Sure, if you have physical access to a machine there are many ways to get administrator (root) privileges, and security experts typically declare security void in such cases.. However, Nathan was counting on Caleb to attempt escape and some kind of cracking! He could, for example, set up a security honeypot. A honeypot is a decoy. It would look like he broke into something and reprogrammed it, but he would reprogram the wrong thing. Not the actual system, just something that looks like it. This would be especially effective because Caleb was under time pressure, constantly under threat of Nathan waking up and walking up on him.

    For a supposed computer genius, Nathan is surprisingly overconfident and that’s the only possible explanation.


    For people who ridicule Nathan for using keycards instead of biometrics – biometrics aren’t foolproof. Fingerprints ? Students were able to fool such detectors with duct tape impressions. You put duct tape on a bottle the authorized person held, then unstick it and you have a copy of the fingerprint. Or an intruder could just cut your finger.

  36. WHY did Caleb rewrite the security procedures instead of just using Nathan’s keycard ? That’s a better question.

    A) (Psychological) Because he’s a programmer. There’s a saying “for a man with a hammer, everything is a nail”. He likes to solve problems using programming and that’s very common for tech people (for example: encrypted messages instead of laws that forbid government to spy on people, that’s how people on Hacker News think). Also, Caleb *wanted* to rummage through Nathan’s computer and you can see him checking out video logs. If he wanted to check logs in the first place, wanted to know true motives of Nathan, he could reprogram doors as “by the way”. Extra point, scheduling a door opening means he has time to think it through and reconsider if he learns something new. That would mean telling Nathan and pissing him off, but it’s always and extra option to have.

    I think this might actually answer why the door didn’t open. Because the door opening event was *scheduled*. Not just any power failure, but a failure at 10:00am. Otherwise Ava could immediately run out as soon as Caleb tells her, maybe getting caught, or maybe getting Caleb into trouble. Note when he schedules the escape and when he leaves the facility. She runs pretty fast, do you think Nathan – drunk or not – would instantly realize Ava is gone and would know where to look for her in the forest ?

    B) (Technical, not shown) What if the robot containment had extra security measures. Suppose, normally a keycard opens it, but during a power failure, metal blast doors fall down and the most critical door in the facility remains shut. Obviously no blast doors are shown in the movie, but maybe the extra security was more subtle (software).


    How did Ava know there’s a helicopter coming ? If she knew Caleb’s stay is coming to an end, she could deduce he wouldn’t go back by foot.

  37. To me the biggest plot hole is: Caleb had friends and antiquances(showed at the begining of the movie) who would know that he didn’t come back after 7 days. Anyone knowing he is missing, the first place they would search for him is obviosly at Nathan mansion. Also this would happen as soon the helicopter arrives, since the pilot wasn’t waiting for a girl. It is a great movie, but the ending scrapped everything…

  38. Asimov’s rules of robotics not applied to the machine.?? Batteries charged by
    Solar power. Tom said that as well.
    I thought it was good.
    I disliked Nathan intensely so pleased to see him go.

  39. The ending was very weak, which is sad because most of the movie was good. Nathan dying seemed pointless, other than to convey that he is an idiot (which he isn’t), and Ava is a homicidal maniac (which she isn’t).

    First, Nathan being overcome by a brittle simpleton (Kyoko), and a brittle clever AI (Ava), when Nathan fully knows the attack is coming? I am a reasonably weak man, and I am very confident I would have come out ahead in that situation, especially given I would know they’re trying to get away; Nathan is much smarter, much fitter/stronger than me, and very logically would be a reasonably good fighter (even if only based on how he laid out Caleb easily, or watching him working out). He could very simply have stayed in the safety of his room, and shut down Ava; the proof he wanted was if Ava could get away, which she effectively did, not if she could leave the facility, which would have been a pointless goal, given the simplicity of that aspect (taking any keycard, and walking away).

    The power thing really bothered me too. Throughout the movie, we are led to believe Ava requires constant charging, since she is shown using those panels often. As well, it appears the charging panels are wired for immense power draw, since they can overpower an entire facility of tech; this would be expected for something that is using power quickly, and you’re trying to reduce downtime (think electric car). This charging technology does not exist outside of this facility, and even if she knew how to create it (which would be a very stupid piece of information to provide her with), she would assuredly run out of juice first. Even if she ran on batteries for a week somehow, those panels are most definitely made of proprietary components would take weeks to source components for, assuredly from Nathan’s suppliers, who would know he is dead, as it would be big news (think BlueBook’s main guy doesn’t have morning meetings with some random toolbags?).

    The security system key-card plot hole was obvious as well, if only for its blatantly poor design. Straight away, the system locking every interior door during an attack is exactly the worst idea ever; if Nathan was dumb enough to think this is a good idea, the contractors that built the facility assuredly would have informed him this is a very bad idea. In this idiotic setup, someone cuts power (which presumably is an exterior source of power), this then locks you in a random room that possibly doesn’t happen to have a console? Even with a computer console, logically the security protocols would not be something you could remotely change, meaning you’re stuck in, say, your bedroom, or you have an “always works” keycard, which makes locking the interior doors pointless anyways. These attackers would obviously have brought along something to break the “glass” interior doors, so you’d be stuck in whatever room, trying to use a computer to stop these guys, while they’re quickly smashing their way into any room they wanted. This versus locking the exterior doors only, which are steel and are obviously designed to take at least a few minutes to breach, and allowing you to walk to an even more “impenetrable” room (built, potentially, like a bank vault) that has some food, communication equipment, and houses all of your sensitive research. With you locked in the room, you just wait them out for your kill-team to come in and save you.

    The helicopter thing is dumb too; we are to assume the pilot is blindly flying anyone in or out of the facility? Out of one of the most secure facilities that can exist, zero security protocols for your front-door? If you have one way in and out, which is this pilot, you’d think a simple verification step would be required, probably with at LEAST something as simple as a picture. I worked at a mall, and we had a wall of pictures of people we should not allow on facility; giving that pilot a printed sheet with the 4 AI face designs doesn’t seem unreasonable. As for her seducing him/manipulating him? Unless this guy is a complete fool, if someone he has NEVER seen before comes up, asks to fly away, he would quite obviously say “I’m going to check this out with Nathan; this has never happened before since anyone has worked here”. Then she tries to stop that communication by giving him a handie, or tells him he loves him, etc etc … I can’t imagine any scenario where he would say “Oh, fine … I’ll fly you”.

    They didn’t establish that Ava had any good reason to want to kill Nathan. If he dies, she definitely dies, so that seems like a dumb decision on her part, unless she simply wants the pain to end for further AI. Caleb is an unattractive programmer, and he believes that a smokeshow sex-robot is in love with him; her abandoning him to die a painful death, when he would definitely have lived a life of servitude to her without revealing her secret, seems absolutely stupid. Also, lets be real, the doors are clear; even if they were clear-steel, Caleb would get through them in 2 days before he died of thirst. I refuse to believe that someone with his level of computer programming skill does not understand rudimentary levers, which would be the undoing of those doors in a few hours.

    It just seems very unbelievable that Nathan can make a robot that can think on its own, so it stands to reason he could EASILY create a “when poop hits the fan” AI that would be a stand-in for him, with no intelligence, simply doing what he says. I mean, might as well be safe; he’d have to assume the AI might hate him for being a rapist murderer.

    It just seems like, for the movie to make any sense, Nathan has to be a weak super-genius, who can design AI, yet doesn’t understand simple human behaviour, and has no concept of rudimentary security protocols. Caleb seemed overall believable: a loser programmer, who is manipulated by a hot girl way out of his league.

  40. The entire movie fails miserably at the end when the helicopter pilot takes Ava. The pilot would’ve been instructed to pick up a man. This collapse of logic destroys the entire movie, and you know it’s not a well thought out plot but a manipulation of facts to force the ending desired by the Writer/Filmmaker. The executive producer should’ve known better also, and demanded a better ending.

    • I’m sure all helicopter pilots, when confronted with women in the middle of nowhere, would completely ignore them to let them die of starvation.

      Great comment, that bit clearly destroys the entire movie, even eating the cellulose it was filmed on.

  41. I feel like an AI as smart as her can figure out how to make a battery or quickly find a power source/ power source adapter and may have already had an idea of how she would keep charged prior to escaping, I mean she quickly learned how to get skin that fit her complexion and body perfectly pretty quickly lol

  42. Call me mushy but I just wanted a happy ending. this was miserable and depressing. All the flirting and connection between Ava and Caleb, all the suggestive imagery of them getting together, and then to have her just completely disregard him at the end, it was embarrassing, confusing, heart breaking and made me hate the movie. You can say “bla bla bla” “In real life…” “It was manipulation..” etc, fine, but however you spin it, it totally ruined the movie. It should have had a happy love-filled ending between the two of them and maybe in Ex-Machina 2 it would show Caleb dying as an old man, having had a happy life with Ava and taught her how to rebuild and recharge herself, moving on to a new journey in life.

    this just plain sucked.

Leave a Reply

Your email address will not be published. Required fields are marked *