There is no doubt that Artificial Intelligence is advancing very rapidly. It can handle the driving responsibility of a car as well as aid doctors in diagnosis. While its capabilities seems endless, there are some weird, interesting and hilarious things that AI system does from time to time.

In this compilation, we are going to write about some funny interesting facts and incidences surrounding the world of AI. From bizarre chatbot conversations to robots with strange habits, these stories remind us that while AI can be impressive, it can also be hilariously unpredictable. So here we go.

Showing Image of a well-known business women on the board of defaulters

In China, facial recognition system is heavily used in many cases in many industries. One such case includes showing the face of the jaywalkers on the bill board in order to encourage people from not making mistake. However, one time, the face of the well-known business women Dong Mingzhu, who heads Gree Electric Appliances, was shown on the Billboard. The Image shown by AI-based survelliance camera made a mistake of capturing Dong’s image from an advertisement covered moving bus. The traffic police of Ningbo announced that they will tweak the AI system so that similar mistake does not occur.

Assumed a bald head as soccer ball

During a local football game played in Scotland, the AI camera got confused by the referee’s bald head and assumed it was the football on numerous occasions causing gaffes. Every time the AI camera adjusted itself to focus on the referee’s bald head thinking of it as the football, the commentator had no option other than to apologize.

The teams playing that match realized the hilarious incident happening at that time and one of the clubs even suggested to the referee to cover his head with a toupe or hat so that the AI Camera doesn’t get deceived due to the bald head.

There is no doubt that the human bald head may appear like a football from certain angles. The incident went viral on the Internet highlighting the inefficiency of the AI Camera at that time.

Underwear to have AI sensors

The underwears of the future may have an interesting feature. According to Myrant Technology, a smart underwear would have micro sensors that would be able to measure sleep quality, stress level, body temperature and other important parameters. The data then would be passed onto the corressponding app which will provide suggestions related to lifestyle changes.

In fact, there could be all sorts of AI-sensors undergarments for both genders if we are to believe an organization like Myant who especializes in making wearables.

You have been exposed to AI way before ChatGPT

ChatGPT was the first noticeable form of AI that has hyped the AI future especially of the Generative AI. But do you know that social media platforms like Facebook, Youtube and Netflix have been using AI for quite a long time.

Those recommendations in your Youtube feed come from AI as the system understands your preferences. The same goes on others platforms because they want to maximize the user stay in order to earn revenue through advertisement.

They may develop their language

In 2017, the research team at Facebook experimented with AI bots to negotiate with each other about owning items that exist only in the virtual world. Something interesting starts to happen after few days as the bots started communicating with each other in a language that was different to human language. In short, the bots had created a new language on their own in order to communicate with each other. The real intent of the bots is still unknown as of today but some speculate that maybe the bots were trying to modify the human language to have successful conversations.

They can create humor but without the comic timing

While AI can create humor for you in various forms, be assured that you cannot rely on it for the comic timing. As you may know, comedy is all about delivering dialogue at the right time. This shows that AI although is good at intelligence it has yet to go a long way in understanding human emotions.

Intelligence

You may think that AI is understanding and has the ability to question beliefs but that is not true. The intelligence of the AI model is as good as the dataset on which it is trained. If that particular dataset has biasedness towards a certain religion, group of people, or towards a belief, it will reflect in its answers.

The biasedness in AI can creep in from training data and algorithmic Design. For e.g., AI systems trained on historical data may give outputs that would reflect the social inequalities.

Generative AI cannot refine itself without Human created content Dataset

As you all know that lots of human-created data from online as well offline sources was used to train Generative AI. Now that new data can be produced by Gen AI, it begs the question whether there would be lesser need of human-created data to train Gen AI. The answer is No! Human-created data is paramount for the betterment of Generative AI.

According to Irving Wladawsky-Berger for MIT Initiative on the Digital Economy opines that humans will always be required to generate new and quality data on which Gen AI model would need to be trained because when models are trained on model-generated data, the results may include defects which are irreversible, a phenomenon popularly known as Model Collapse.

Taking everything Literally

AI doesn’t operate on human-like understanding. Its complete foundation is built entirely on data patterns and algorithms, unlike humans who can understand context, text, and subtext. Understanding idioms and metaphors may take some time and training for the AI.

Can create much better jokes than humans

According to a study released by the University of Southern California, AI is more capable of creating jokes much better than the humans. The model that was used for creating jokes was of ChatGPT 3.5 version. This may come as a surprise especially since AI models are typically considered incapable of understanding human emotions that too in a joke.

The above news may instill fear in artists and in the entertainment industry who may believe that AI will take up their role, however it’s safe to say that novel ideas and creativity may be something that AI will never achieve.

Power hungry

Artificial Intelligence is making waves across industries with its incredible results but that surely is not without major expenses. As per a report published in the journal Joule, the likes of ChatGPT built by OpenAI consume electricity which is a whopping 17000 times more than the average household in the US.

According to a report by Goldman Sachs, the demand for data center power will increase by 160% by 2030 because of the AI revolution.

While running AI at the moment requires significant computing power, there are reports that AI can mitigate nearly 10% of global GHG emissions by 2030.

Conclusion

These are some of the funniest, interesting facts and noteworthy incidences of Artificial Intelligence Gaffe. AI is revolutionizing the world in a way that we could have never imagined but it is also true that AI is a source of amusing and intriguing facts. There will be areas where its full implementation may cause a gaffe or an error if remains unsupervised by a human being.

But just like anything, one thing that we can be certain is that AI is also an evolving technology and its ultra-rapid advancement would mean that problems like hallucination, biasedness, and other errors may get eliminated sooner than we expect.