On our social media channels, we show interesting facts or figures on the topic of artificial intelligence at regular intervals.
Adversarial Attack
AI systems are far from infallible. With the increasing presence of AI, research is more focused on so-called “Adversarial Attacks” – a method of fooling AI systems through manipulated data input. The best-known example is a stop sign covered with a few stickers that has caused autonomous vehicles to interpret it as a sign for a speed limit. Or the image of a panda that was classified by the AI as a gibbon due to the addition of carefully designed visual noise. There is no single solution for protection yet, but possible methods and strategies are being researched. It is an evolving field and one that is definitely worth keeping an eye on.
Game over. From the iconic chess match between Deep Blue and Garry Kasparov to Watson beating the top two all-time champions on Jeopardy – these historic matches demonstrated the incredible potential of AI to process vast amounts of information and outperform humans in certain domains. Back then it was a sensation, today we are far beyond that and AI is being felt in every area of the gaming universe.
Siri, Alexa, Cortana & Co. – the power of AI is female, at least when it comes to voice assistants. There are various reasons for this. The systems available to companies like Google at the time were trained on more female data than male data. Additionally, several studies have shown that female voices are perceived as more sympathetic and pleasant. While many voice assistants can now be switched to a male voice, the default setting remains a female voice, which 90 percent of users are satisfied with – a clear majority! Do you share this opinion?
He did not say that?! In a 2018 video, former President Obama called Trump a “complete dipshit” – one of the most well-known examples of a so-called “deepfake”. At the time, the video was quickly exposed as a fake. However, with the continuous development of AI, the quality of deepfakes is also improving. A good example of this is the deceptively real videos of “fake” Tom Cruise that went viral on TikTok. But beyond all the entertainment, deepfake technology is also a tool for fraud and misinformation and can become a danger to society. In the future, any statement could be put into any politician’s mouth – would you believe it?
Before Siri and Alexa, there was ELIZA. Admittedly, ELIZA’s functioning doesn’t have much to do with the AI that underlies digital assistants like Siri or Alexa nowadays. The program, with which users could chat in text mode, was not particularly intelligent. Nevertheless, ELIZA is considered an important milestone in the field of AI.
In the 1960s, researcher Joseph Weizenbaum developed ELIZA at the Massachusetts Institute of Technology (MIT). He pursued the idea that ELIZA would respond to users’ problems like a therapist and help them cope with their emotional challenges. Although ELIZA only returns messages based on recognized keywords, some users almost had the illusion that they were speaking with a real person.
Source: www.nextpit.de / Weizenbaum, J. (1966). ELIZA – A Computer Program For the Study of Natural Language Communication Between Man and Machine. Communications of the ACM, 9(1), 36-45.
Autonomous cars
A charming idea, but at the same time controversial: getting in and sitting back while your car chauffeurs you autonomously to your destination. According to a survey, 45% of drivers doubt the reliability of self-driving cars and fear hacker attacks. However, autonomous vehicles would bring a significant improvement to road traffic, as human error is the cause of 90% of accidents. Nevertheless, we’ll have to wait a little longer; a recent study predicts that fully autonomous cars won’t be available in large numbers until 2040.
Sounds familiar? As soon as you enter a website, a chatbot offers its help – 24/7. This offer is being taken up more and more frequently. Especially when it comes to quickly answering a simple question, 69% of consumers contact the chatbot. So, in the near future, the AI tool could replace the classic “FAQ” pages. By the way, the satisfaction rate is 87.5%, which indicates an improvement in the customer experience.
