From time to time, we may rely on AI to help us. But, before we hand over the keys of society to our automaton overlords, let’s recognise that there’s another side to it that’s all too human. 

Let’s look at AI’s failed attempts in 2022 to replace us flesh-and-bones types!

Meta’s BlenderBot blunder

Conversational AIs have had a long history of goof-ups since the beginning. Adding to the list, this year, ‘BlenderBot3’ made several false and jarring statements in its conversations with the public. 

In August, Meta released its chatbot, BlenderBot3, to the public to collect feedback within a few days. First, it described Meta CEO as “too creepy and manipulative”. Secondly, it claimed anti-semitic conspiracy theories as truth and then stated that it still believed that Trump was the US President.

Chess robot breaks a child’s finger in Russia

A chess-playing robot, unsettled by the quick responses of a seven-year-old boy, broke his finger during a Moscow Open match in July.

“The robot broke the child’s finger”, Sergey Lazarev, president of the Moscow Chess Federation, told the TASS news agency following the incident, adding that the machine had played many previous exhibitions without upset. “This is, of course, bad.”

Deepfake fiascos

Several deepfake videos have gone viral over the past two years. President Obama used profane terms to describe President Trump. Mark Zuckerberg admitted that Facebook’s true goal is to manipulate and exploit its users, Bill Hader morphing into Al Pacino on a late-night talk show.

Since the trend caught up, the subset of machine learning has had arguments around it. But, even with the potential it holds to make a positive impact, incidents highlighting the dangers of deepfakes constantly surface on the internet every now and then. 

• Patrick Hillman

The Binance chief strategy officer knew something was off when he found six emails in his inbox from clients about recent video calls with investors he had allegedly participated in.

Hillman realised that someone had deep-faked his image and voice well enough to hold 20-minute calls trying to convince his company’s clients to turn over their Bitcoin for scams. “The clients I was able to connect with shared with me links to fake LinkedIn and Telegram profiles claiming to be me inviting them to various meetings to talk about different listing opportunities,” he said.

• Kate Isaac

Imagine your face being edited into a pornographic video without your consent and shared on the internet. Kate Isaac revealed the horror of it happening to her.

Scrolling through her Twitter feed one evening, she stumbled across a video. Isaac had been deep-faked. Someone had used AI to manipulate her face into someone else’s — in this case, an adult film actress.

• President Volodymyr Zelenskiy

Another video purporting to show Ukrainian President Volodymyr Zelensky publicly capitulating to Russian demands drew widespread ridicule. Still, experts said it could be a sign of sophisticated deceptions.

The video showed an ashen-faced Zelenskiy speaking from the podium, urging his fellow citizens to surrender to the Russian invaders.

Internet users immediately flagged the mismatch between the skin tone on Zelenskiy’s neck and face, the odd accent in the video, and the pixelation around his head. A Facebook official later said the company removed the footage from its platform.

​​Virtual Rapper Fired

Capitol Records dropped an AI rapper and offered their “deepest apologies to the Black community” after the internet denounced it for the racist stereotypes in the song by using the N-word and being portrayed as being beaten by a police officer.

Created in 2019, the AI rapper ‘FN Meka’ is given the appearance of a Black male cyborg. According to Martini, his songs are performed by an anonymous Black man, but an AI generates the music and lyrics that analyse popular music.

In Other News – 

• LaMDA is sentient(?)

The question baffled the science and tech community this entire summer. Announced at Google I/O in 2021, the conversational AI managed to stay the talk of the internet even a year later.

At the end of the debate, there is no indication that LaMDA was truly sentient. There’s no formal way to prove sentience these days, but a chatbot that ticks all questions listed above is a good start. In 2022, however, LaMDA is far from achieving that.