Pintrest’s Response to AI Slop Inundation

Also, DeepSeek claims theoretical 545% profit margin

Author

Jack Lajoie
March 03, 2025 • read time ~ 9 minutes

⚡️ Headlines

🤖 AI

Parents and teachers: how are you introducing AI to younger children? - The Guardian seeks insights from parents and educators on methods and motivations for teaching AI to younger children. [The Guardian]

Apple's Artificial Intelligence Efforts Reach a Make-or-Break Point - Apple's attempt to deliver a more advanced, conversational version of Siri has encountered significant delays, with the most modern version now expected no sooner than iOS 20, possibly arriving in 2027. [Bloomberg]

Crossing the uncanny valley of conversational voice - Sesame aims to achieve "voice presence" by developing AI companions capable of engaging in genuine dialogue that builds confidence and trust over time. [Sesame]

OpenAI plans to bring Sora's video generator to ChatGPT - OpenAI intends to integrate its AI video generation tool, Sora, directly into ChatGPT, enhancing the chatbot's capabilities. [TechCrunch]

SoftBank in talks to borrow $16 billion to fund AI, The Information reports - SoftBank Group is negotiating to borrow $16 billion to invest in artificial intelligence, with plans to potentially borrow an additional $8 billion in early 2026. [Reuters]

Samsung Broadens AI Phone Lineup With $300 5G Galaxy Models - Samsung introduces a new lineup of 5G Galaxy phones starting at $299.99, featuring AI enhancements in photo editing and other functionalities. [Bloomberg]

China's Honor announces $10 billion investment in AI devices - Smartphone maker Honor plans to invest $10 billion over the next five years to develop AI across its range of devices, including PCs, tablets, and wearables. [Reuters]

China Tells Its AI Leaders to Avoid U.S. Travel Over Security Concerns - Chinese authorities are advising AI entrepreneurs and researchers to limit travel to the U.S. due to concerns over national security and the risk of divulging sensitive information. [The Wall Street Journal]

🦾 Emerging Tech

Donald Trump Names XRP, SOL, ADA, BTC and ETH as Part of U.S. Crypto Reserve - President Trump announced the creation of a U.S. strategic cryptocurrency reserve, including assets like XRP, Solana, Cardano, Bitcoin, and Ethereum, leading to significant market value increases for these digital assets. [CoinDesk]

Tesla Sets Sights on Waymo, Uber in California Ride-Hail Bid - Tesla is seeking approval to offer ride-hailing services in California, aiming to compete with companies like Waymo and Uber by utilizing its autonomous Full Self-Driving vehicles. [Bloomberg]

Google's Taara Hopes to Usher in a New Era of Internet Powered by Light - Alphabet's project Taara aims to deliver high-speed internet using laser beams, offering higher bandwidth at lower costs compared to traditional methods. [Wired]

SEC's 'Demolition' of Crypto Enforcement Met With Cheers as Well as Jeers - The SEC has dismissed or paused several cases against crypto companies, signaling a shift in regulatory approach that has been met with mixed reactions. [Bloomberg]

🔬 Research

Anthropic partners with U.S. National Labs for first 1,000 Scientist AI Jam - Anthropic collaborates with U.S. National Laboratories to evaluate frontier AI models on scientific research and national security applications during the first-ever 1,000 Scientist AI Jam. [Anthropic]

🎱 Random

Amazon's Delivery Drones Are Grounded. The Birds and Dogs of This Texas Town Are Grateful - In College Station, Texas, residents experienced significant disruption from Amazon's Prime Air drones, leading to community backlash and the eventual grounding of the drones, restoring peace and allowing wildlife to return. [Wired]

🔌 Plug-Into-This

Pinterest, once a hub for aspirational lifestyle imagery, is now inundated with AI-generated content, often referred to as "slop," overshadowing genuine human-created posts. This influx spans categories like home decor, fashion, and recipes, leading to user frustration over the platform's diminished authenticity.

  • Prevalence of AI Content: Searches such as "healthy recipe ideas" reveal that at least half of the top results are AI-generated, with minimal labeling to distinguish them from authentic content.

  • Misleading AI Personas: Many AI-generated posts link to dubious blogs featuring fictitious authors with AI-created profiles, misleading users seeking genuine content.

  • User Dissatisfaction: Long-time users express frustration, noting that the abundance of AI content has made it challenging to find real-world examples and diminished the platform's usefulness.

  • SEO Exploitation: Content creators exploit AI to produce mass content, aiming to manipulate search algorithms and generate ad revenue, further diluting content quality.

  • Platform Response: In reaction to these challenges, Pinterest plans to implement more transparent AI labeling on posts to help users identify synthetic content.

📉 So it begins…well, honestly, social media platforms were already backlogged with low quality content. The pace at which it can be made though, has gotten way faster. It was always looking like the end result for platforms which structure incentives into a game of “who can get the most views at the least cost”.

Chinese AI startup DeepSeek has disclosed theoretical profit margins for its V3 and R1 models, suggesting a cost-profit ratio of up to 545% per day.

  • Operational Costs: DeepSeek estimates that renting Nvidia's H800 chips at $2 per hour results in a daily inference cost of $87,072.

  • Projected Revenues: The company projects daily revenues of $562,027 from its AI models, leading to the stated 545% cost-profit ratio.

  • Actual Earnings Caveat: DeepSeek cautions that actual revenues are lower due to factors like free web and app access, varied model costs, and reduced developer fees during off-peak hours.

  • Impact on AI Industry: These claims have raised concerns among investors about the capital expenditures of U.S. AI firms, potentially affecting companies like Nvidia, Broadcom, and Marvell Technology.

  • Comparison with U.S. Counterparts: DeepSeek's reported lower training costs, under $6 million, contrast with higher expenditures by U.S. firms such as OpenAI.

💰 ”So ChatGPT is ripping us off?” is the general response so far. That shouldn’t be a surprise, though. Even if we temper these numbers against the idea that Chinese based DeepSeek may still be running a destabilization/devaluation campaign against US companies, it’s useful to see under the hood into what kind of profit margins are actually possible with AI products.

Dario Amodei, CEO of AI startup Anthropic, recently appeared on The New York Times' "Hard Fork" podcast to discuss the dual nature of artificial intelligence, emphasizing both its vast potential benefits and significant risks.

  • Public Awareness: Amodei expressed concern that the public is not fully aware of AI's risks, predicting that awareness will increase within the next two years, potentially causing a shock.

  • Potential Misuse: He highlighted the danger of AI being misused by malicious actors, which could pose threats to millions, and noted that AI could enable autocratic regimes to enhance their control.

  • National Security: Amodei stressed the importance of liberal democracies maintaining a competitive edge in AI to prevent abuses of power and protect national security.

  • Balancing Risks and Benefits: He believes it's possible to address AI risks without hindering its benefits, but this requires nuanced discussions and careful implementation.

  • AI Safety: While acknowledging the challenges in controlling AI models, Amodei remains optimistic about developing effective safety measures in the future.

🦺 Amodei's insights underscore the urgency for proactive measures to ensure AI technologies are developed and used responsibly, balancing innovation with safety.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse