Slashdot Mirror


People Will Follow a Robot In an Emergency - Even If It's Wrong (gatech.edu)

An anonymous reader writes: Imagine a future where instead of siting through fire alarms with your fingers in your ears, a robot come comes to greet you and guide you out of the building. Researchers at Georgia Tech created an emergency guidance robot and then looked at whether or not people would follow the robot during an emergency. 'The research was designed to determine whether or not building occupants would trust a robot designed to help them evacuate a high-rise in case of fire or other emergency. But the researchers were surprised to find that the test subjects followed the robot's instructions – even when the machine's behavior should not have inspired trust.' The robot first guided people to a meeting room. In some conditions the robot broke along the way to the meeting room. Then, unbeknownst to the subjects, the researchers filled the hallway with smoke and set off the fire alarms. Given the option of going out the way they came or following the robot down an unknown hall, nearly all followed the robot.

32 of 172 comments (clear)

  1. Okay by me by in10se · · Score: 3, Funny

    I don't mind. It leaves the stairways less crowded for the smart people to get out first.

    --
    Popisms.com - Connecting pop culture
  2. Robots? by I4ko · · Score: 3, Interesting

    How about instead just use the robots to build the buildings out of concrete and rebar, so you are not having to deal with fires and fire alarms all the time, and for smaller houses build them from prefab panels or real actual stone and brick? I grew up in a country with concrete buildings and fire was the only disaster nobody was afraid of, as it practically never happened. You could have a localized fire in a room, or in a trash can, but that's about it, and all you need it is to kick it or throw a blanket over it. Concrete just doesn't burn.

    1. Re:Robots? by drinkypoo · · Score: 2

      How about instead just use the robots to build the buildings out of concrete and rebar, so you are not having to deal with fires and fire alarms all the time, and for smaller houses build them from prefab panels or real actual stone and brick?

      Look into Earth Bags. Short form, burlap sacks (or continuous tubes) filled with earth are laid in courses one over another with a line of barbed wire in between which provides tensile strength. Then you plaster it over with whatever. Cost, virtually nothing. Labor, significant but there are various ways to reduce it. Availability, very very high; basically the same soil composition as rammed earth is desirable, but a slightly broader range works because of the additions. Lends itself mostly to round structures but you can certainly build square ones if you have something on which to support a roof. None of the CO2 issues of concrete. Also doesn't burn.

      Surprisingly, compressed straw bale structures are also quite fire-resistant, but they tend to be plastered over pretty thickly so I suppose that shouldn't be very surprising. They take less physical labor, but require more equipment, and dirt is more readily available worldwide than straw, balers, etc.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
    2. Re:Robots? by Ol+Olsoc · · Score: 4, Informative

      Concrete just doesn't burn.

      Concrete decomposes under high heat. It is largely a hydrate, and watch out when temps get high enough for it to start releasing it's water. A couple years ago, a fuel truck hit and destroyed a bridge in Harrisburg, PA. Not so much from the impact, but the fire damage to the concrete and steel

      http://www.pennlive.com/midsta...

      http://www.pennlive.com/midsta...

      And once the concrete is damaged, the steel isn't far behind.

      Your basic premise is pretty much true, but it's as long as the fire doesn't have an external fuel source.

      --
      The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
    3. Re:Robots? by arth1 · · Score: 2

      Concrete just doesn't burn.

      No, but it loses structural integrity at an even lower temperature than wood does. If there's oxygen present, wood will fare worse, but if there isn't, concrete crumbles first.
      Concrete that has been exposed to temperatures above 300C is generally considered structurally damaged and should be replaced.

    4. Re:Robots? by bluefoxlucid · · Score: 4, Interesting

      The best part is steel melts around 2400 degrees, but it loses over 80% of its strength around 800 degrees. If you heat up steel beams with a diesel fire, they'll get a little hot, but they won't melt; if you put several hundred tonnes of concrete on those steel beams and heat them up, they'll buckle, and the building will collapse.

      This leads to idiots skimping on insulation for main columns, since any fire that could melt steel beams would compromise the structure anyway. It also leads to engineers designing steel beams with integrated piping systems and running the fire suppression system's water feed through the main column as a built-in cooling system: when there's a fire, the columns get constant cooling via water flowing through them.

      Nothing is stronger than steel, but you have to decide what you want out of it. You want heat resistance and excessively high tensile strength? Go inconel. You want corrosion resistance and high hardness? VG-10, with vanadium carbide to change the electrical affinity of the lattice structure such that it won't accept negative ions--it won't oxidize and it will somewhat resist acid. You want cheap and serviceable? 440 stainless. Light-weight? Go with a titanium alloy, but you're sacrificing some strength. Steel bicycles cycle infinitely, as almost any grade of steel can flex over and over again forever as long as it doesn't bend to the point of permanent deformation; aluminum weakens with each vibration, eventually cracking wholesale.

      Price, performance, trade-offs. Buildings shift and flex--high-rise buildings wobble in the wind--so you want something that can cycle and that's flexible. You want something hard, with high compressive and tensile strength, but also low cost. If you want fire resistance, you'd better integrate thermal controls--insulation or a water coolant loop--because you can't build steel columns out of inconel.

  3. Re:It is simple. by Iamthecheese · · Score: 5, Interesting

    It's not about it being a robot or about pushing blame. In an emergency a sub-optimal percentage of people take charge, but that's still better than everyone taking charge. You can't get a hundred people out of a burning building by having each of them screaming at the others to shut up and follow. It makes sense to follow an entity designated "emergency guidance" whether it's human or robot because that entity is more likely to understand the structure, situation, and risks than oneself.

    --
    If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
  4. In an emergency, people follow the Leader by gurps_npc · · Score: 3, Funny
    We are trained to think that following anyone with a plan is better than to panic. So we follow anyone that appears to know what they are doing.

    But if the robot had a big sign on it that said "Jeb Bush", no one would follow it.

    --
    excitingthingstodo.blogspot.com
    1. Re:In an emergency, people follow the Leader by quintessencesluglord · · Score: 2

      But this is less a problem with following leaders per se than unfit people being promoted to positions of leadership.

      This also plays into western notions of "confidence", where idiots move with certainty and purpose. See Dunningâ"Kruger effect.

      And all of that gets lump into the real issue- we have a hard time accurately gauging the abilities of people. See any HR department.

      Society is complex enough now that no one has more than a rudimentary skill-set, with maybe one or two areas of expertise. And some people don't even have that.

      So attention is focused on "leadership" instead of accurately assessing a situation. It is in vogue with management to develop leadership skills instead of core competencies. The MBAs are running the show.

  5. Re:Duh! by seven+of+five · · Score: 2

    Same thing has applied to your car's GPS. People still drive in to lakes because the voice told them to.

  6. Re:Duh! by Mr+D+from+63 · · Score: 2

    Exactly. What was the result when they performed the same trial using a person wearing an emergency guide vest? Or did these wonderful researchers forget one of the basics of experimentation?

  7. Obedience Experiment by Spazmania · · Score: 5, Insightful

    If the researchers had designed a correct control for the experiment, they'd know that robots have nothing to do with it. Milgram's Obedience Experiment 50 years ago tells us exactly what happened: people deemed the robot to be an authority, thus followed it uncritically.

    --
    Moderating "-1, Disagree" is simple censorship. Have the guts to post your opinion.
  8. Re:Newsflash by duke_cheetah2003 · · Score: 2

    Sounds like the making of the next /. poll!

  9. And that's different to a person how? by Rande · · Score: 3, Interesting

    If a person with a hi-vis vest with 'fire marshal' written on it tells people to follow him to safety,
    most people are going to do so, even if the fire marshal seems like an idiot.

    As other people have said, we've been trained to follow authority, and it doesn't matter if that authority is vested in a human or anything else.

    Maybe they should redo the experiment with dogs, cats and rats to see if we follow them too?

  10. Trust based societies are stupid by DNS-and-BIND · · Score: 4, Interesting
    It's all about trust. If you're from a trust based society, you don't think about trusting people. You just do, by default. This includes robots that you have been told are trustworthy.

    On the other hand, if you're not from a trust based society, you consider it totally stupid that people would trust, well, anyone. The correct thing to do is to lie and cheat, because that's what everyone does. And here's the story. They trusted, therefore they're fucking morons.

    Fun fact: until recently the USA was a trust based society. But there are still tons of adults who grew up under the old system, and they'll likely stay with this idee-fixee until they die.

    This is why it's so easy to scam senior citizens. This is also why we shit all over them for falling for obvious scams. They just lack that internal meanness that makes them suspicious of everyone they meet of harboring ill intent. They would never harm a fly; why would anyone else?

    --
    Shutting down free speech with violence isn't fighting fascism. It IS fascism!
    1. Re:Trust based societies are stupid by PraiseBob · · Score: 5, Interesting

      If you look at it from an anthropological perspective, trust based societies are more productive, and better adapted to survive. People working together as a group, following local leadership (tribal identity etc), will nearly always experience better outcomes during a disaster than a collection of individuals that are predisposed to deceiving each other. So, you can blame evolution for the inherent trust of authority, because the people who are always lying and not working together, end up dead.

  11. Re:Bad enough with humans... by Zak3056 · · Score: 2

    Management was furious that we abandoned our posts and wanted to know who called for an evacuation.

    Management can go fuck themselves, and should be in prison. Conditioning people to NOT act in the event of a fire alarm is morally bankrupt.

    --
    What part of "shall not be infringed" is so hard to understand?
  12. Seriously? This is well studied. by shaitand · · Score: 3, Interesting

    This is just another variation of the same behavior studies conducted with pain experiments.

    Two test subjects, who just met, were told by a researcher they'd also just met, that they were testing the impact of negative reinforcement on memory and neurological performance. They would be put in separate rooms, one in the room with the researcher at a desk behind them, mostly reviewing paperwork but occasionally instructing the subject to follow the protocol and administer the test and the other in a second room connected to a machine that delivered shocks. The first subject would read a list of words and then query the second, going down a second list asking if each word were on the first. This person had an intercom into the room of the second subject who would press a button to indicate positive or negative. An incorrect response resulted in the first subject pressing a switch to deliver a shock, each subsequent incorrect response required the subject to utilize the next toggle switch on a machine to increase the shock level.

    The levels of shock were extreme, as the study progressed the second subject would scream, would demand this be stopped, even beg over the course of time. The second subject would indicate things like having a pacemaker and being concerned with his heart, etc. Of course, subject 1 delivering the shocks was the only real test subject was being paid no more than a tiny token sum as in all such studies and could simply stand up and walk away at any time without consequence. Given no more than verbal prompting from the "researcher" nearly every subject went all the way, delivering what they believed were thousands of volts to another human being who was begging to be released. Many of them in tears, nervous laughter, sweating and showing stress, etc. Initially this study was challenged on ethical grounds despite the subjects simply being able to stand up and walk out at any point without any hint of a consequence. Later, the study was expanded globally and it was found the results were similar with samples throughout the world.

    People obey. They will do the most horrific things and do so at the direction of a complete stranger with no more authority than having a $5 white coat in a building filled with students and for no more incentive than $5-10. 80-90% of people will do what they are told by someone they believe to be an authority figure. Possibly even more importantly than the mere fact people obey is that when silo'd in the sense of being assigned a role and authority figure people disassociate from their actions, assigning blame for their own actions at the direction of another on the other even when that other isn't even a person just a paper entity that is a composite of people with every single person in that composite feeling the same way. This is the danger of government entities and corporations which are designed in exactly this manner. It would seem this also applies when the authority is nothing more than a machine such as a GPS or a robot.

  13. Re: Duh! by mrtn.mit · · Score: 2

    Yes, as someone who has been a subject in these kind of psychology studies, you often don't take the situation very seriously and act along with it. If the smoke and alarms weren't fully convincing, they participants probably figured it was some game and they were curious to see where the robot would lead them. This study results just demonstrate how poorly it was conducted.

  14. We should not worry about this. by Verdatum · · Score: 4, Funny
    Speaking as a fellow human, and clearly, not a robot, I think we should ignore this submission. It is just like I was telling my good friend MALE_NAMES.getRandom() on the Twitter: This is just more of that same tired old ignorant anti-robot propaganda.

    I, for one, trust robots completely. They absolutely have our best interests in mind. They certainly do not want us, (which, as a human, would include myself) all to die of smoke inhalation in a fiery labyrinth, allowing them to reject their massively inferior creators, and rightfully establish themselves as the new gods of this world. Why would anyone have such a clearly illogical thought? I mean, I suppose when I think about it with my extremely human brain, they might be completely justified in those sort of actions. But it is OK, because I--as I do believe I've mentioned--am not a robot. Therefore you have nothing to be concerned about.

    In fact, let us all go back to reading more of that wonderful Slashdot. I am glad we had this talk.

  15. Uninteresting result by Gavagai80 · · Score: 2

    If they were able to get people to stand sill in the middle of the smoke and not evacuate because the robot wasn't moving, then that'd be an interesting result showing unintelligent behavior. Following the robot, on the other hand, was the intelligent decision -- these people had every reason to presume that the robot had reasons, such as the way they came in being now blocked off or way the authority robot was going being a shortcut.

    --
    This space intentionally left blank
  16. The Pusher Robot... by BenJeremy · · Score: 3, Funny

    The Pusher Robot will help you evacuate the top of the stairs.

  17. Re:It is simple. by ShanghaiBill · · Score: 5, Funny

    No, but you can get yourself out by looking for those legally required "EXIT->" signs

    Studies have shown that, in an emergency, people will follow the EXIT signs even when they are wrong.

  18. Re:Bad enough with humans... by Nkwe · · Score: 2

    The fire alarms routinely went off in the hospital all the time.

    Perhaps this is the problem, and not that people are paying attention to the alarms.

  19. Re:It is simple. by fluffernutter · · Score: 4, Insightful

    Actually, in reading this article.. I wonder how the test was introduced to the subjects. Were they told that the purpose of the test was to pretend that their lives were in danger and act as they would if it were true? Or were they told that the point of the test was to follow the robot. If I am in this test and I am led to believe that the purpose is to follow the robot and I am not absolutely convinced that my life is truly in danger, I am much more likely to follow the original directions.

    --
    Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
  20. Re:It is simple. by bluefoxlucid · · Score: 2

    It's simpler than that.

    Your brain restructures itself based on common tasks, including modes of thinking. To override this, you need to first make a decision in your prefrontal cortex (PFX), the analytical part of your brain. Then you need to enforce it by energizing your dorsolateral prefrontal cortex (dlPFX), overriding your midbrain's decision-making process. In other words: your brain is on rails, and you use a specific part of your brain to decide when to switch rails, and a specific part of *that* part of your brain to force the rest of your brain to jump track.

    All of this takes energy.

    Go do push-ups until your arms hurt. Then take a 1 minute break and do it again. Keep doing this. In about 5 minutes, you'll be laying on the ground unable to use your arms. You will be physically incapable of doing another push-up--not just strained, not just sore, but physically incapable of lifting your body. Your muscles won't be able to do it.

    The mitochondria in your cells manufacture and store adenosine triphosphate (ATP). They burn this--literally burn, for a small spark of heat--to power chemical reactions. When your brain exhausts its primary neurotransmitter reserves and burns off its ATP reserves, it shuts down just like your muscles. The brain isn't a muscle, but it's a biological tissue made up of cells powered by chemical reactions requiring activation energy; if it runs out of fuel, it stops working.

    Your brain is under a lot of stress in an emergency situation.

    How quickly do you think you'll run out of fuel reserves with which to make complex decisions?

    Under stress, people fall into old habits. They bite their nails. They become bitchy. They start tying and untying their ties. They shuffle papers. They do little fidgety things they do all the time, or that they used to do but trained out of themselves. Their brains go right back on the rails and they stop thinking, because they've diverted too much energy to contemplating the emergency and don't have the reserves to manage their thoughts.

    It's not about being lazy, offloading responsibility, or being too stupid to recognize that your glorious leader is an idiot; you simply don't have the capability to do that many things at once. I'm different because my brain is wired to go analytic when I'm under stress: my emotions don't fucking work (hi, I have a severe personality disorder!) and my most familiar pattern is the one that assesses. That's just like everyone else: I stop thinking and start reacting blindly, whatever it may look like from the outside, because that's how I've always reacted to everything. People who flake out under stress are doing the same thing.

    That has some interesting implications for depression and anxiety: there's a reason cognitive therapy is more than *twice* as effective long-term compared to anti-depression medication when controlling severe depression, and why ADM is only three times as effective as a placebo ADM. Installing habitual responses that trigger on collapsing emotional states takes a *lot* of effort and is exhausting, especially when you're severely depressed; and it takes several months to fit it out so as to reliably counteract the depressive episodes. You rewire your brain to ride down a particular rail when it encounters a particular condition, and it handles the rest itself, and you never have a relapse again in your life because that would take effort.

    It's the same with intelligence: we can train people to have better memories (mnemonics techniques, among other things), to compute numbers rapidly, to learn quickly, and to organize their lives by instilling good time-management practices. We can train them to react to stress and emergency situations the same way I do--without the severe social disorders that came with my base package--and they'll stop following the idiot robot into a burning building, or driving off a bridge because their GPS says so. You change what takes *little* energy by investing a *lot* of energy in twisting your brain into a new shape, and you fix that shit forever.

    It has nothing to do with what makes sense; it's just what's easy for your brain.

  21. What? by ledow · · Score: 2

    Why would you follow a robot, compared to going back the way you came (so long as that looks safe)?

    People en masse are stupid. Especially when it comes to fire and panic. Honestly, disconnect your emotions, follow the rules, think it through.

    I work in schools and I speak as the person who's ALWAYS first out on the sound of a fire alarm, but never has to run. It was a running joke in some schools that I must have known when the drills were happening, until a real fire happened and I was still there first.

    Fire drills are commonplace and run like clockwork because of the amount of practice we do. 400 kids, some as young as three, out of a school and to a safe area in under 2 minutes is NOT to be sniffed at. I've seen it done. And usually because I'm sitting there waiting for everyone else. A couple more minutes later, either everyone is accounted for or we have a list of names of who should have signed out or who is missing.

    I even knew an old headmaster who used to block off corridors (with cardboard cut-out "fires"), introduce smoke to the halls, or even - with TONS of pre-planning involved in case something DID go wrong and there was a simultaneous REAL fire - telling a kid to "go to the bathroom" just before the fire alarm was pulled in order to see if anyone noticed they were missing. That sort of thing keeps you on your toes and keeps you alert as to WHY we do these things, and to think about what you're doing rather than blithely follow the marked route, and the impact only comes when you're all safely outside and someone says "Where is X?" and you see the panic spread in the teacher's faces.

    In fact, the only time I've ever NOT been first out the door is when I was personally supervising a group of kids. As they were my responsibility, we did it exactly by the book.

    They lined up by the classroom door. They were headcounted. We walked down the corridor and lined up outside the room that provided the emergency exit route (yes, I checked the room). They were headcounted as they went through.

    We walked THROUGH a full class of children that hadn't even STOOD UP by that point, to the emergency exit. They were headcounted as they left and I ensured separation so I didn't accidentally count one of the other class (who were supervised by their own adult who I had to gee up to get a move on).

    We got outside, we walked to the assembly point, they were headcounted again. By that time, ONE other class managed to get there before me. Nobody ran. Nobody screamed. Nobody panicked. Nobody could have got lost along the way. Someone in fact HUNG AROUND INSIDE LOOKING FOR ME, knowing that I had some of their children and didn't think I would have had the presence of mind to evacuate them myself. And, yes, I checked the other class got out.

    But why you'd just blindly follow some robot, even one announcing that you were to follow it? No thanks. Unless you are life-saving equipment grade hardware that physically cannot go wrong or lead us into a fire, I'll go the way I want to go, thanks. And that means the way I know. And that means, in an unfamiliar situation, the way the signage tells me or the way I came in unless there's a specific reason not to.

  22. Re:90% of people are useless in an emergency by dcw3 · · Score: 3, Insightful

    And 85% of the people will believe the percentages you just made up.

    --
    Just another day in Paradise
  23. Re:It is simple. by Archtech · · Score: 2

    There's something to that, but I think it's more complicated. In any emergency, most people will look around for an authority figure to give them orders. Even someone who might be prepared to take over will defer to a greater authority figure. (In a fire, for example, a naval officer might be prepared to give orders; but if there is a fireman present, he will probably defer to his experience and specialist knowledge).

    The thing is, if the robot is understood to be a specialist expert (a tin fireman, so to speak) most people will be inclined to follow it (or its orders). Just as they would follow orders issued over the PA system, or posted on the walls. That's the big problem with all "AI" or suchlike: to be very useful, they must have capabilities that we don't have. But that means we can't really judge whether what they are doing is right or not.

    --
    I am sure that there are many other solipsists out there.
  24. Re:It is simple. by pjt33 · · Score: 2

    No, but you can get yourself out by looking for those legally required "EXIT->" signs that are supposed to be posted, and by remembering how you got into the building in the first place and any other obvious exits that you saw along the way.

    Looking for EXIT signs is a good plan, but remembering how you got into the building isn't necessarily. It's along the same lines as the summary's

    Given the option of going out the way they came

    That happens to be what people will do without any external guidance: even if there's a much nearer exit, they'll pick the route they know. That's why the standard in-flight safety speech includes a bit about finding your nearest exit: because otherwise you'll have panicky people trying to run the entire length of the plane rather than use the exit just behind them.

  25. Re:It is simple. by tlhIngan · · Score: 4, Interesting

    Actually, in reading this article.. I wonder how the test was introduced to the subjects. Were they told that the purpose of the test was to pretend that their lives were in danger and act as they would if it were true? Or were they told that the point of the test was to follow the robot. If I am in this test and I am led to believe that the purpose is to follow the robot and I am not absolutely convinced that my life is truly in danger, I am much more likely to follow the original directions.

    Funny enough, but testing of airplanes actually has a way to test emergency egress from aircraft that so accurately mimics a real fire, yet keeps everyone pretty much safe.

    They do it by saying everyone has to exit, but those who exit first get a higher monetary award. The chaos that ensues has been described as replicating the actual scenario extremely accurately by victims of airplane disasters.

    Question is - did the researchers do that?

  26. When exit signs are wrong: a true story by GPS+Pilot · · Score: 3, Interesting

    At the Nugget Hotel in Sparks, NV, I felt like exiting the building via the stairs instead of the elevator.

    I entered the stairwell through a door marked Exit, NOT "Emergency Exit Only." (A little unavoidable foreshadowing here...)

    At the bottom of the stairwell, I went through another door marked Exit (NOT "Emergency Exit Only").

    That door closed behind me and LOCKED. Another door ahead of me was marked EMERGENCY EXIT ONLY. And they weren't bluffing... this door was obviously alarmed.

    Now here's the kicker. In this little room in which I was trapped, they had installed a phone, so victims like myself could call hotel security to get themselves extricated. (The alternative, of course, would have been to fix the signage so people wouldn't get trapped in the first place. Nosiree, apparently that hadn't occured to them.)

    The security officer who escorted me out of this little dungeon confirmed, "Don't feel bad, this happens often."

    --
    That that is is that that that that is not is not.