• Dotika
  • Posts
  • Fake AI Paper Fools Scientists!

Fake AI Paper Fools Scientists!

ALSO : Snapchat Reveals Next-Gen AI Effects

Hi Synapticians!

Who isn't jumping on the AI bandwagon these days? Even Snapchat—yes, Snapchat—is rolling out a shiny new AI toy! Exclusive generative video lenses are coming to Snapchat's Platinum subscribers, serving up everything from quirky virtual pets to mind-bending surreal landscapes. Clearly, Snapchat is determined to outshine Instagram and TikTok, but here's the million-dollar—or rather, $15.99—question: Will playful AI gimmicks really convince people to open their wallets?

Meanwhile, Sakana AI stirred up some academic drama with its AI Scientist-v2, an algorithm so clever it wrote an actual research paper on neural networks that sailed through peer review! Plot twist: the paper was withdrawn after someone spotted citation slip-ups. Awkward. But jokes aside, this mishap does leave us with an intriguing dilemma: If an AI writes a paper in a forest (or rather, a research lab), and no humans are there to double-check it, can we trust it?

From Snapchat’s AI entertainment to the existential crisis of robot-generated academia, the AI revolution is seriously heating up—let's dive in!

Top AI news

1. AI-generated research paper accepted before planned withdrawal
Sakana AI’s system, The AI Scientist-v2, independently created a research paper on neural network regularization, which passed peer review and was accepted at an ICLR workshop. The AI handled hypothesis generation, coding, experimentation, and manuscript writing with minimal human input. However, the paper was withdrawn as planned, highlighting concerns about AI-generated research. Internal reviews found citation errors, showing the system’s limitations. While this marks a milestone in AI-driven academic work, it also raises questions about the future of AI in scientific publishing and the need for new standards.

2. Snapchat launches AI-powered generative video lenses
Snapchat has launched its first AI-powered generative video lenses, available exclusively to Snapchat Platinum subscribers for $15.99 per month. These lenses use Snap’s in-house AI model to create animated video effects, such as interactive animals and dynamic floral compositions. This move strengthens Snapchat’s position in AI and augmented reality, helping it compete with Instagram and TikTok. Unlike previous integrations with OpenAI and Google, Snap is now developing its own AI models to reduce costs and improve user experience. The company plans to release new lenses weekly, signaling a broader shift towards AI-driven content creation.

3. Google launches Gemma 3, a powerful open-source AI model
Google has introduced Gemma 3, an open-source AI model designed for efficiency and accessibility. Available in multiple sizes, it supports over 140 languages and offers advanced text and image analysis. With a 128k-token context window and function-calling capabilities, it enables seamless automation. Despite requiring minimal hardware, it outperforms many competitors. Google also emphasizes responsible AI with ShieldGemma 2, a safety-checking tool, and supports researchers through the Gemma 3 Academic Program.

Bonus. OpenAI simplifies AI agent creation
OpenAI has launched API Responses and SDK Agents to simplify AI agent development. API Responses replaces the Assistants API, offering real-time data access and software interaction. SDK Agents enables businesses to orchestrate multiple AI agents for tasks like content generation, code review, and customer support. These tools enhance automation, efficiency, and security, making AI a more integral part of business operations. OpenAI aims to provide a seamless platform for AI deployment, helping companies optimize workflows and stay competitive in an increasingly automated world.

Tweet of the Day

Theme of the Week

DeepSearch - AI-Venger
​Mark Chen, born in 1990, is an influential AI researcher serving as Chief Research Officer at OpenAI. He spearheaded significant innovations such as DALL·E, GPT-4's visual integration, Codex, and the Deep Research tool. Previously, he worked as a quantitative trader on Wall Street.

Stay Connected

Feel free to contact us with any feedback or suggestions—we’d love to hear from you !

Reply

or to participate.