Subscribe
Sign in
LiarLiar.ai — Combining psychology with computer vision
Introducing LiarLiar.ai, a groundbreaking leap in AI lie detection technology. Designed to discern truthfulness and identify potential deception, LiarLiar operates in real-time, combining proven psychological methods with advanced computer vision.
225
Replies
@nastassia_k Hello Nastassia, thank you! We are not using the Paul Ekman technique, yet. However, we use a neural network that's trained on millions of images for different emotions, while for the other type of micro-expressions, we are actively tracking more than 400 facial landmarks, and we have basically hard-coded the cues we're looking for such as change in eye gaze, blinking, lip compression, etc.
@asenlevov intersting. Do you have a monthly plan ? If not can I try it out with a video eg ww.youtube.com/watch?v=qO161zSEHho
@asenlevov Congratulations on the launch, LiarLiar.ai! 🚀 Here's to a future of enhanced security and trust! 🌐🤖
@saurabh_parmar Thanks, shoot me an email at support@liarliar.ai with the video you want to see :)
@ian9k Hi Ian, we have really put a lot of late nights into building this product, and it's the furthest thing from a scam. Is it a perfect software? No. Does it have a lot of bugs and issues? Yes. But, does it work as intended? Yes. We have a long way to go with this project, and this is yet the beginning, so get ready to see our digital communications change slowly :)
@asenlv do you know how many people have had their lives ruined by faulty lie detector tests? people still think polygraph exams are real. you admitted *in this thread* the product doesn't work on huge swaths of the population, do you think that's going to stop anyone from using it and acting like it's perfect? None of this throat clearing about accuracy is on the site.
Combining psychology with computer vision for lie detection is definitely a unique approach. I'm curious to know how accurate the system is and if it has been tested extensively. Also, are there any plans to integrate LiarLiar.ai with existing security systems for enhanced fraud detection?
@jorge_aguilar_avendano Hello Jorge, thanks for the spot-on question. We have done about two dozen experiments so far, and we have reached a detection accuracy of close to 75%, while the standard polygraph with wires and all would sit at around 85%. Of course, it has all come down to calibration and reinforcement learning at this stage for us, and our next big milestone would be to further improve our technology, make more experiments, and then possibly aim at higher-security use cases. For now, the lighter ones might include human resources, sales talks, and everyday communications. We know for a fact technologies like our have been used by some border control agencies around the world, but haven't shown much promise due to privacy concerns.
@asenlevov, it is amazing what technology can do today! It is inspiring to see how you have combined psychology and computer vision to create LiarLiar.ai. We can only imagine how exciting it must be for you to have made it available to the public. Well done!
Oh wow! You've finally launched, watched over your "Coming soon" for a while guys 😁
As I wrote before, just took the service for a spin, and it feels like a peek into the future of truth detection.
Back in a while, I worked hard on face emotion recognition and many more (in stickerface.io and in Capture https://www.producthunt.com/products/capture-6, look at your project and see that technologies have stepped to the next level. Keep it up guys, good luck! 🔥
question: the header of PH says lifetime access, but I guess it does nothing. there are no promo codes or something?
@mattisssa Haha thanks for following us, Artem! I'm glad you see some progress in tech - let's always see that and have the world become prettier, better and cleaner :) The lifetime deal is still site-wide and available for everyone to get. Planning to switch to subscription by the end of the year.
It's truly something we will need, especially with all the recent advancements in AI and beyond. I also enjoyed seeing the ingenious ideas of use, as André J mentioned. 😄 Great product, and congrats on the launch! 👏
This is truly a launch I'm super excited about. congrats on building this. Would love to get it some time soon 🔥
Bro this is scary, but i would love to see some case studies using this
like for detecting stress & overworking employees on call, and then offering them some holidays to relax
This is very cool! Definitely going to try it out! Congratulations on your launch! 🚀🚀🚀
Casually wondering if LiarLiar was measured against neurodivergent reactions, especially autistic people. We aren't going to have the same responses to stimuli on our faces. So, chances are, if this data-set didn't include that...it might not work accurately for folks who process things differently.
@kiraieigh Hello there, thank you for the appropriate and spot-on comment. People with mental and even body disorders would surely have varying reactions, and our tool wouldn't be a good measurement in such cases. We have done no such experiments so far. We know for a fact that psychopaths/sociopaths would pass any polygraph test flawless, because they would truly believe what they are saying, so in these types of cases, the standard tools or even the new-age tools like ours wouldn't be applicable.
Congratulations to the LiarLiar.ai team! This is such a revolutionary idea; the application of AI in lie detection is absolutely groundbreaking. I admire the novel concept behind real-time truth discernment and deception identification. My suggestion would be to consider introducing a credibility scale or spectrum in the output, offering a more nuanced understanding instead of a binary lie/truth scenario. Wishing you every success with LiarLiar.ai!
@asenlv
Wow, this sounds like a fascinating venture! 🚀 How does your tool differentiate between someone being naturally sensitive to a question and intentionally being deceitful? I'm intrigued to understand how it navigates the nuances of human emotions and responses.
@asenlv
Wow, this sounds like a fascinating venture! 🚀 How does your tool differentiate between someone being naturally sensitive to a question and intentionally being deceitful? I'm intrigued to understand how it navigates the nuances of human emotions and responses.
@aashish_peepra It doesn’t differentiate natural sensitiveness, but it does learn and adapt to the specific person’s behavior, and it works by detecting fluctuations from that normalized behavior.
Big shoutout for the product! What's the next significant milestone or feature you're eyeing in development?
@nikita_detkov Thanks for asking, Nikita! We're planning to do more refinements to the existing application over the next couple of months, and then go multi-platform with it, including an API service.
@sentry_co Hi Andre, you can send the video to support@liarliar.ai, and I will process it for you gladly :)
@asenlevov Awesome! thanks so much. I think this is so exciting! Imagine this as an Apple TV app analysing live tv, or as a mobile app to verify if people are telling the truth 😅 Black mirror vibes! So this is a famous left wing politician from Norway who has admitted that he lied in this interview: (full disclosure I have nothing against him or his party, I'm in the middle politically, This is truly to test if this service works, as his body language is all over the place 😅) https://www.nrk.no/video/moxnes-...
@sentry_co My pleasure, Andre! I hope I don't come off as the villain in this Black Mirror episode, haha. Here is the analyzed video: https://youtu.be/AVM5FXIgtTc
@asenlevov Thanks! That's really impressive stuff! I had no clue about all these signs that tell if your lying or not! Only problem is that the very same politicians can now practice lying with the same tool that will catch them. 😂
@sentry_co It would take too much practice - they have more lucrative things to do surely Haha
PolygrAI