“I thought I was done covering Elon Musk for a while after sending out TechScape last week. But then the news editor reached out, asking me to monitor Musk’s Twitter feed.
So, I ended up closely tracking the world’s most influential social media enthusiast, and it left me mentally exhausted.
His shortest break from posting came on Saturday night when he logged off after retweeting a meme that compared London’s Metropolitan police to the Nazi SS. He was back online just four and a half hours later, retweeting a crypto influencer’s complaint about the jail sentences given to Britons attending protests.
Despite my familiarity with Musk’s online presence—split between promoting his businesses, Tesla and SpaceX, sharing low-brow nerd humor, and engaging in right-wing political commentary—I was still taken aback by what I found.
Following Musk in real-time highlighted how his chaotic online persona has evolved as he’s shifted further to the right. His promotion of Tesla has taken on a culture war tone, with the Cybertruck being marketed as a tool to help defeat Democrats in the upcoming U.S. presidential election. The humor he shares is increasingly tinged with frustration that the world doesn’t see him as the coolest person alive. Meanwhile, his political posts are becoming more extreme.
Musk’s engagement with the unrest in the UK appears to have drawn him closer to far-right figures. Recently, he tweeted at Lauren Southern, a far-right Canadian internet personality banned from entering the UK over her Islamophobic views. Musk isn’t just tweeting; he’s also financially supporting her, sending around £5 a month through Twitter’s subscription feature. He also retweeted a post from the co-leader of Britain First, a far-right political group. Initially, this might have seemed like Musk was unaware of the company he was keeping, but two weeks on, it’s clear that these are now his allies.
Switching gears, there’s a stark contrast between how scientific press releases and papers can present findings, as seen in a recent case from the AI world. A press release from the University of Bath claimed, “AI poses no existential threat to humanity.” The release suggested that large language models (LLMs) are safe, predictable, and controllable because they lack the ability to learn new skills without explicit instructions.
However, the paper by Lu et al. offers a more nuanced view. It argues that so-called “emergent” abilities of LLMs—tasks they perform without being explicitly trained—aren’t as mysterious as they seem. Instead, these abilities result from in-context learning, model memory, and linguistic knowledge. The paper suggests that while LLMs may excel in some tasks and falter in others, their abilities should not be overestimated. This finding is intriguing, though it falls short of proving that AI will never pose an existential risk. Still, for those looking for reassurance, it suggests that an AI apocalypse isn’t imminent.
On a related note, Nvidia is facing a federal lawsuit over its use of YouTube content to train its AI systems. The lawsuit, filed by YouTube creator David Millette, accuses Nvidia of “unjust enrichment and unfair competition” for allegedly scraping his videos without permission. Nvidia’s methods reportedly involved evading YouTube’s detection to download vast amounts of content for its Cosmos AI software. This case is notable because Nvidia has been more secretive about its data sources compared to other AI companies, which have been openly dismissive of copyright constraints.
Unlike other AI companies that have faced similar lawsuits, Nvidia wasn’t as transparent about the sources of its training data. In contrast, companies like Stable Diffusion openly used data from the open-source LAION dataset, leading to legal challenges over copyright infringement.
However, not every AI company faces the same hurdles. Google, for example, benefits from the fact that site owners often can’t afford to block it from using their content to train AI, as doing so would also remove them from search results entirely.”