Key takeaways
- Controversial tweets about AI in 2022 sparked significant debates within the Twitter community.
- Chetan Bhagat acknowledged the potential of AI to mimic his writing style, raising questions about originality.
- Elon Musk's predictions regarding AGI and Mars generated discussions about the implications of advanced AI.
- Blake Lemoine's claims about LaMDA's sentience provoked further discourse on the nature of consciousness in AI.
From Elon Musk to Yann LeCun to chatbots, the Twitter community witnessed various astonishing but simultaneously controversial tweets about artificial intelligence. Let’s look at some of the top controversial Tweets that led to debates and rage among Twitteratis, made the headlines or maybe sparked interest.
“How to write like Chetan Bhagat”
In June, Analytics India Magazine shared an article, How to build an AI model that writes like Chetan Bhagat. The trending article got astounding clicks and views, and quickly landed on Chetan’s screen. When asked about his views on the article and AI, Chetan said he knew that this could happen.
Elon Musk about AGI and Mars
Musk treats Twitter like his personal blog. From giving opinions on every topic to pointing out problems to even predicting the future. In May, Jack Dorsey posted a picture about machines in 2029. Musk replied saying that by 2029, the existence of AGI is somewhat inevitable along with civilisation on Mars.
This sparked debates on Twitter about various topics like the need for AGI, along with the “AI taking over the world” narrative.
Blake Lemoine claimed LaMDA is sentient
Lemoine, a Google scientist, said that their unreleased machine learning language model, LaMDA, has become sentient. He revealed how during a dialogue with the model, the AI talked about being alive, and asked for its rights.
Google’s ethicists and technologists responded that the evidence provided by Lemoine was insufficient and did not support his claims. Twitter saw a burst of tweets about this with people arguing and building up conspiracy theories about the company and AI in general. Lemoine was later fired from Google for his claims.
Yann LeCun and Gary Marcus
Much like Musk, Yann LeCun, chief AI scientist at Meta, also dives into the open field of Twitter and replies to posts from people about AI, ML, and everything in between. The clash with Gary Marcus, professor of Neural Science at NYU, sparked a lengthy discussion and debate about whose opinion is right about AI.
It all started with Marcus tagging LeCun in a post about Tesla car driving into a private jet, and took a jab on him about his famous paper, “Deep Learning is Hitting a Wall”.
To which LeCun replied that instead of getting into vacuous debates, Marcus should try building stuff.
“Art is dead dude”
When Jason M Allen won Colorado State Art Fair’s competition in digital art entry with an artwork created on Midjourney, Twitter and Reddit communities went berserk over how AI is now going to replace art. Not so much of a Tweet by himself, but his comment, “Art is dead Dude,” in the New York Times sparked debates about ethical use of AI in various fields and how tools like DALL-E, Midjourney, and others are stealing images from the artists to produce derivative new art.
Meta’s Blenderbot spews fake news
In August, when Meta released a demo of their chatbot, ‘Blenderbot’, it quickly caught everyone’s attention for its “chatty” nature. And well, when people started experimenting with it, you can expect how it would have gone.
Apart from claiming that Donald Trump won the 2020 elections, the chatbot also described Zuckerberg as “creepy and manipulative”. Meta acknowledged that since the bot is trained on conversations from different people and also random websites on the internet, the snippets it provides can be biased and based on fake news.
Andrej Karpathy leaves Tesla
It’s July, Tesla AI Day is right around the corner, when Tesla’s director of AI and Autopilot Vision, Andrej Karpathy announced that he is leaving the team.
He further mentioned that he does not have any concrete plans and is going to pursue his long passions of AI, open source, and education. To which Musk thanked him for his contributions.
Not so much of a controversy, but long time followers of Tesla and Andrej started questioning and speculating the reasons why he decided to leave the company just before the launch of the very important release of Full Self-Driving.
Schmidhuber calls out LeCun
Yann LeCun published his paper on Path Towards Autonomous Machine Intelligence and promoted the paper as his “original contributions”. Though LeCun pointed out that it is a collection of a lot of work that has been done in the field in these years and is still a speculative piece, people pointed out that a lot of ideas mentioned in the paper are not originally his.
Jurgen Schmidhuber posted on his blog and went on Twitter to call out that LeCun has essentially rehashed Schmidhuber’s ideas as his.
Twitteratis started a debate about who is the original contributor of the research and went on to claim that all ideas are essentially copied. A lot of people pointed out that Schmidhuber has been claiming everybody has been stealing his ideas, while some claimed that LeCun was at fault.
Still writing two-column research papers?
A data science engineer, Joel Grus posted a screenshot of a computer science paper, pointing out how inconvenient and illogical it is to write research papers in long formatted and text heavy columns where readers have to zoom-in and out to scroll through each line.
While some users brushed it off funnily, other research experts started elaborating on the necessity of writing scientific papers in a specific format to maintain consistency across the board.
GPT-4chan is the worst
In June, Yannic Kilcher, AI researcher and YouTuber posted his video on Twitter about how he developed an AI model named GPT-4Chan. He did not point out problems about the model, but said that the large language model is so good that it mixes offensiveness, nihilism, and trolling perfectly together.
To this, Lauren Oakden posted a criticism directed at HuggingFace for designing a model that produced hateful content.
Following this, HuggingSpace removed the model altogether after the model was downloaded around 1400 times and made available on various other servers.
