Welcome to the second edition of The Neural Net…
Happy Valentine’s Day, everyone! We’ve got love, drama, and AI all wrapped up in today’s newsletter - because what’s more romantic than a little tech talk?
In this issue:
Leaders debate AI in the City of Love
Artists attack AI art auction
Anthropic details the typical use case of it’s chatbot Claude
Meta lays off low performers, begins AI hiring frenzy
AI struggles to summarize the news
Athletes flock to startup ScorePlay
From Paris, With Love
For Valentine’s Day, we’re taking you to the city of love - except instead of candlelit dinners on the Seine, it’s world leaders and tech giants whispering sweet nothings about AI policy. The Paris AI Action Summit, held on February 10-11, brought together leaders and tech executives from over 100 countries to discuss the future of AI. It was packed with big declarations, bold investments, and, of course, a little drama.
To start, Vice President JD Vance declared that AI will “never replace humans”, but should “make us more productive, more prosperous and more free.” He then moved on to proclaim that “American AI will not be co-opted into a tool for authoritarian censorship.” Seems like there is no love lost between Vance and fact checking - I mean censorship. Meanwhile, French President Emmanuel Macron announced a €109 billion investment plan, a clear response to the U.S.’s Stargate initiative. Nothing like a little rivalry amongst lovers to start a spending spree.
The biggest topic of contention was what type of romance should be had with AI - a passionate no-holds-barred whirlwind that ends with a wedding in Vegas, or a “date for 30 years before committing” snooze-fest. The Americans, who have never made a rash decision in their entire history, lean toward the former, with Sam Altman warning of the need to widen the mindset around AI and not focus so much on risk. The Europeans are more the types to pack 4 sets of clothes for a 1 day trip (just in case), and prefer the latter with a higher emphasis on regulation and safety over innovation. As a result, the U.S. and U.K. declined to sign a declaration endorsed by around 60 countries at the summit, including China, India, and Germany, which aimed to ensure AI technology remains "safe, secure, and trustworthy". To paraphrase JD Vance - safety was so 2022, now it’s all about opportunity.
Whether this summit was a love story or a messy breakup remains to be seen, but one thing’s for sure—Paris wasn’t just about romance this week. AI took center stage, and the future is swiping right.
Heard in the Server Room
Quick hits roundup of AI news
If Sotheby’s can sell a banana for over $6 million, it’s clear the definition of art is expanding and Christie's wants in on the action. Christie’s is hosting its first auction dedicated entirely to AI-generated art, featuring 20 pieces estimated between $100 and $1.7 million. However, the event has sparked backlash, with over 3,700 artists signing a petition arguing that AI models exploit copyrighted works without permission. In an ironic twist, the backlash is also providing the opportunity to train the models into tortured artists whose art nobody understands.
Anthropic has launched the Anthropic Economic Index to analyze AI's impact on labor markets. They performed the study with anonymized data from millions of interactions on their model Claude.ai. The report finds that AI is widely used in fields like software development and technical writing. Additionally, Anthropic found that 57% of AI usage dealt with augmentation (enhancing human work) vs 43% automation (directly performing human tasks). At least that’s what the AI probably used to summarize these findings wants you to think.
Meta is ramping up its hiring of machine-learning engineers to build and train AI models, while cutting about 5% of its workforce by letting go of underperforming employees. In other words, the company is firing people to hire people who will eventually help automate their jobs out of existence. CEO Mark Zuckerberg emphasized the need to enhance performance management by swiftly removing low performers and replenishing these positions with qualified candidates. You know, the kind that wear sandals and T-shirts to work.
AI News Summaries: Fact or Fiction?
The BBC is on a mission to fact check generative AI, and let’s just say... the results aren’t great. In a recent study, they put four of the leading AI assistants - ChatGPT, Copilot, Gemini, and Perplexity - to the test, having them summarize news articles before real journalists checked their work. The verdict? 51% of summaries had major issues, and 19% flat-out got key facts wrong - we’re talking numbers, dates, and other hard facts. The takeaway from BBC’s study? “AI assistants cannot currently be relied upon to provide accurate news and they risk misleading the audience.” Not exactly a glowing review.
This whole study might have been inspired by the now-infamous Apple AI news summaries - the push notifications that were supposed to make catching up on the news easier on your phone. Instead, Apple had to pause the feature in January after users noticed the AI wasn’t just getting things wrong - it was making things weird. The summaries weren’t just misleading; they were often bizarre, overly dramatic, or straight-up clickbait. In one instance, an AI-generated headline mistakenly claimed Luigi Mangione shot himself - he did shoot someone, just not himself. Talk about missing the mark! Maybe Apple was secretly beta-testing a game of Two Truths and a Lie? Some of the best (or worst) examples include:
1. source
2. source
3. source
Spot the lie—test your knowledge from the news above:
Littler was prematurely declared the winner of the largest darts tournament while it was still ongoing
Netanyahu was very much not arrested, much to the ICC’s chagrin
Nadal is Spanish, not Brazilian…
Start Me Up - ScorePlay
Pull the curtain back on the hottest companies and startups in the AI space
ScorePlay is an AI-powered sports media company that aims to use AI to deliver personalized sports content, enhancing how fans engage with games, highlights, and statistics. Founded in 2021 by Victorien Tixier and Xavier Green, ScorePlay’s platform automatically tags and organizes content, enabling teams to quickly distribute it to broadcasters, sponsors, and athletes.
It has raised over $13 million in series A funding, bringing its total funding secured to around $20 million. With an investor pool that looks like a who’s who in the sports world (Giannis Antetokounmpo, Nico Rosberg, Dominic Thiem, Kevin Durant, Eli Manning, and many more), this startup is ready to bring home the hardware.
That’s all folks! Show some love to your fellow AI humans today, and we’ll catch you in our next edition.
Like what you read? Help us grow the community by sharing The Neural Net!