Examples of Failure in Artificial Intelligence

Artificial intelligence is ground breaking and at times, still is very popular and mind blowing. We often hear stories of efficiency, automation and intelligence.

But AI is not always the perfect. As every success story has a messy failures story- a situation where something did not go well according to the plans.

Okay, enough with the preamble, now let’s deliver the goods. Here are some examples of Failure in AI over past few years. Scroll it down to read.

Microsoft AI Chat bot learns some Unbecoming Language

AI chat bot has become a norm in social media other websites. Facebook has a really good one built into Messenger and it is regarded as the powerful business tool for providing customer service and starting up an conversation. But AI chat bot are not perfect, as exemplified by Microsoft’s AI chat bot, which went by the name of ‘Tay’.

It was released in March 2016 and was used by the twitter users. Tay was programmed to do casual and a natural conversation in the typical common language which lasted only for 24 hours. Later it got manipulated by a grope of trolls and they were using this program for a very sexiest and racist statements or comments.

The researcher at Microsoft, had to seek apology for not seeing the future insight.

Athlete or Felon?

Amazon has a project named as Recognition. It is based on AI facial recognition software. It is used to marketed to police agencies in investigations. It is suppose enforce law to all possible suspects. The problem is that it is not accurate at all.

At least 27 of these athletes were false and were matched with mug shots. This included three-time Super Bowl Champion Duron Harmon of New England.

It is not at all a good luck.

Users find flaws in Apple’s Face ID

Apple is always coming with new and cutting edge technology. They have standard mobile and smartphones industry for many years. They almost they all the things right but sometimes you mistakes may occur.

After the launching of i Phone X, it invested a lot of dollars to front-facing facial recognition system that has replaced the finger print system of accessing the smart phone. The AI was able to recognise the face without glasses, make-up and many more. The problem occurred when Apple claimed that Face ID cannot be spoofed by masks or other techniques.

This was just reminder that often bold claims can shatter you down to earth.

Voice Spoofing AI Software Cons CEO

Deep fakes are becoming a serious issues now-a-days. Hackers have found new ways to adopt fake voices, pictures and even videos. And in certain cases the effect is dangerous.

There was a case when a German patent company got a call from his boss to transfer a certain huge amount of money to an Hungarian supplier. The only problem with the request that it was not the boss over the call. It was an AI based software made to mimic the voice of boss.

Is AI perfect? As the stories or examples provided in this article show that NO! AI is not perfect, but it is really very powerful and ever-improving. Yes on both side it is true.

So as we are looking forward to next three to five years, we shall always remember that AI is ever-progressing. And to get where we want to be, we have to deal with little bit of friction on the way. It comes with the territory.

-Nikita

Categories: News