Post navigation

100 thoughts on “Why Is It So Hard to Make a Realistic Twitter Bot?

  1. Is it sad that I can usually tell when Olivia is hosting just by looking at the Like bar before the video even loads?

  2. People went crazy when Google's Duplex made its debut and demanded it identify itself as an AI. People should demand the same transparency from social media platforms.

  3. Bots so stupid a 3 year old would see through them, also russian bots that you can never see through…. Something is wrong here 😛

  4. You're pretty hard to get ahold of! Do you have time to talk? …Well I won't take up much of your time… Would you say that you live in a building, or a home? …Haha And are you afraid that I might be a robot? …That makes sense. Now can I get you to confirm your social security number? …Oh yes that's what it says here. Well that's all the questions I had, so I'll get going and I think you for your time. …You too.

  5. I've observed some polarization schemes. Like a bot with talk trash about something in order for people to rally and defend the subject effectively improving the subjects perception.

  6. its simple. nothing online is real. the end. its all bots, kids, and Indians paid a penny per post. then a few actual people log in and think these are all people like them completely distorting reality giving people something to rail against or make them feel good to agree.its really creepy when you realize this is pretty much the whole of the internet and people are being manipulated

  7. This is exactly the wrong question to ask. The correct question is why do these companies claim people who aren't bot are (this has happened to me on facebook). It's down to political, specifically left wing bias. They just can't believe ANYONE would have a problem with the tactics and results of politically left policies. No only idiots don't want an authoritarian state to censor everyone's free speech away.

  8. Wrong , why does twitter make me do a test to prove im not a bot , when i comment on liberal stories in negative light

  9. You find them all over the place in youtube comment sections, at least on certain types of videos (political, alternative science/history/etc, anything to do with any sort of suppressed information). So obnoxious.

  10. Because Twitter keeps deprecating APIs for business reasons. Only 1% of developers used them anyway…

  11. ok, so if you teach bots words, but not all the words, then program it to name something, could it do it? like tail and horse, but not teach it what a ponytail is, would it figure it out?

  12. Years ago I told someone some of these things and they accused me of having spent too much time in my room and having become tin-foil hatted.

    😐

  13. I was half expecting this to segue into an ad for Brilliant. A course to learn how to build your own Twitter bot troll army.

  14. Hi, love your channel! Im curious about something that other people would be curious about as well:
    We have exemple of all soft of sex behaviors with mamals, and I supose that is some exception among fishes too. But is there any bird that is not monogamous? Os there any lizard that is polygamous, or at last form a couple for some signifative time?
    Thanks!!

  15. There are a lot of humans who are incapable of any dialogue. Do you think bots could have gone to the next step and created a human shell to fake us?

  16. if you're not sure if it's a human not try asking it if a shoe box is bigger then Mount Everest?

    I would just code it so that every time it's asked a question out of its depth it just replies to the person with " how would I know, if you're so curious YOU should Google it" or "how would I know, there's this thing called, Google ever heard of it?" Or "that's a fascinating question too bad I'm too lazy to bother answering seriously" and other slight variations upon that way it sounds like a real human all the time cuz that's something a real human would say and it's useful in a million situations

  17. The reason you can’t find good bots is because they are too good to know that they are bots
    We only notice the bad bots

  18. they could programme bots and other "AI" assistants like google now or siri to learn. i did a quick test: i asked google if a shoebox was bigger than mount Everest. it gave me the web results for 3 mountains taller then everest. then I asked the average size of a shoebox, it gave me 5x7x2 inches, and i asked how big Everest was, it gave me 8848 m. than i asked it again if the shoebox was bigger, it gave me the same web result as before. if the bot could scan the previous questions and answers it gave, analyse the question and make a simple comparison, it could simply give the obvious answer, therefore learn.

  19. Because people who use twitter are sub human mongoloids that have already begun the process of human de-evolution.

    If you spend more than 10 seconds on twitter you need to be committed.

  20. My favorite kind of bot is the one that is programmed to respond to accusations that it is a bot; typically by saying things like "nuh uh, you're a bot!" Like seriously? That's so convincing! I must be the bot!!!

  21. someone's probably beat me to this but just in case: AI-Complete doesn't mean you'd have to solve all of AI to get there, it means solving this problem is equivalent to (or stronger than) solving all other AI problems. Completeness is a concept from the area of study known as Computational Complexity. It's most commonly used to refer to NP-Completeness. (In that case it's about whether a problem can be solved in non-deterministic polynomial time.) We talk about problems as being NP-Complete if we can map them onto* other NP-Complete problems. Therefore, if we can solve one, we can solve them all as we have mappings between them (and therefore can use the solution from one to create the solution to all the others).

    *I always mix up which direction the reduction goes. Someone please correct me if I have the mapping backwards.

  22. If people or groups are trying to deceive humans or affect human behavior, using a generative adversarial neural network might eventually appear, especially if it can compete against itself. Such a method created an unbeatable AI chessmaster who learned in a matter of 4 hours. How many times have you believed something told to you without proof? Maybe just enough to put doubt in your mind. Presidential campaigns have been won using such tactics. Bush vs McCain in South Carolina by affecting antiblack voters with a story that McCain's adopted African daughter was born out of wedlock. By the time the truth came out it was too late.

  23. the way every sentence ends on this super 'low' semi-growl is grinding my brain. I wish I could articulate this better. I'm so sorry but I have to go.

  24. I have to chat with my company's clients sometimes as part of my job, and it always amazes me when partway through a conversation someone will ask if I'm really human. I don't think many people realize just difficult it is for a computer program to fake being human.

  25. Frankly, bots have already reached parity with the value and intellect of social twitter though in fairness a cat walking across a keyboard could do the same.

  26. A while ago I posted a request for some to tell this chick to stop making me sick with her arm and hand shaking. She has gone from making me vomit from motion sickness to just making me queazy. I’m not trying to say she sucks. She doesn’t. But she needs to use those hand movements a little less like salt. A dash here a dash there. Pepe le pew is great so is Hank. I’m a big fan of this channel. But every time I see this chick I want to sit her down on a chair tie her arms behind a back and tell her to start again. Regardless much love scishow.

  27. So if we take a consensus by studying responses, number of views, likes, etc. we need to consider this new type of "stuffing the ballot box".

  28. This is interesting, I've been spending a lot of time on Twitter lately but despite knowing there's an epidemic of bots I didn't consider how I might have been fooled by them on my own timeline already.

  29. scientists figure out how to make a AI that understands human emotion home intelligence first. Year 2025 is getting close yet no effective chat bots…..

Leave a Reply

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