Uncensored version of DeepSeek R1

Hi everyone,

NYT (the one that sued OpenAI for copyright infringement) is now enabling its employees to use those same tools to write stories - even they can’t get around the massive benefits of GenAI for workplace productivity.

Also, DeepSeek R1 now has an uncensored version, thanks to Perplexity.

Let’s get right into it.

In this issue:

• 🤿Deep Dive: Perplexity tackles China bias and censorship

• 🤝 In Partnership: You won’t believe how easy outbound can be

• 🤿Deep Dive: Grok 3 enhances X network’s AI capabilities

• 🖼️AI Art: Examples of great and trending AI art

• 🤿Deep Dive: New York Times adopts AI with human oversight

• ⚒️Tool Snapshots: Tools for AI, no-code, and productivity

• 📰Top News: News on AI, no-code, and productivity

🤿 DEEP DIVE

Enhancing DeepSeek-R1 for Unbiased AI Responses

Intelligence: Perplexity has undertaken a significant post-training effort to enhance its DeepSeek-R1 language model, aiming to eliminate biases and censorship issues, particularly around sensitive topics related to China.

• Method used:

  - Human experts identified around 300 censored topics from the CCP.

  - A multilingual censorship classifier was developed to facilitate the collection of relevant user prompts.

  - The process generated a dataset comprising 40,000 multilingual prompts, carefully curated to exclude personally identifiable information (PII).

• R1 was post-trained on the curated censorship dataset using Nvidia's NeMo 2.0 framework, designed to effectively remove biases while preserving the model's core reasoning and academic performance.

• A diverse evaluation set of over 1,000 examples was curated to test the model’s ability to handle sensitive topics. This involved human oversight and specialized evaluators to validate response quality.

• Evaluations showed that the post-trained model maintained its performance on par with the original R1, proving that the enhancements did not compromise its reasoning abilities.

🤝IN PARTNERSHIP WITH ARTISAN

Hire Ava, the AI SDR & Get Meetings on Autopilot

Ava automates your entire outbound demand generation process, including:

• Intent-Driven Lead Discovery

• High Quality Emails with Waterfall Personalization

• Follow-Up Management

Free up your sales team to focus on high-value interactions and closing deals, while Ava handles the time-consuming tasks.

Book a demo to see how Ava can 10x your outbound.

🤿 DEEP DIVE

 Elon Musk's xAI Launches Grok 3, a Powerful New AI Model Family

Intelligence: On Monday, Elon Musk's AI company xAI introduced Grok 3, a new AI model family with enhanced capabilities for the social network X, promising advanced image analysis and simulated reasoning features.

• Grok 3 is powered by a significant computing infrastructure at xAI's Memphis data center, utilizing around 200,000 GPUs, and boasts 10 times the computing power of its predecessor, Grok 2.

• Musk suggested that Grok could embody his perspective, with a notable example where Grok criticized traditional media as "garbage," a claim that diverges from standard AI outputs and raises concerns about bias.

• AI researcher Andrej Karpathy highlighted Grok 3’s performance, noting its quick ascent to near "state of the art" capability within just a year of development.

• xAI will offer various plans for Grok 3 access, including a $50 monthly subscription for early access and a new $30/month SuperGrok option with unlimited image generation.

• The Grok 3 family includes several models, such as Grok 3 Reasoning and Grok 3 mini Reasoning, which feature simulated reasoning for improved problem-solving in subjects like mathematics and science.

🖼️ AI ART

Examples of great and trending AI art

Created with Midjourney by dontrun_withscissors, this collection reimagines Africa's iconic animals in the icy wilderness of Antarctica.

🤿 DEEP DIVE

 The New York Times Integrates AI Tools in Newsroom Operations

Intelligence: The New York Times has approved the use of artificial intelligence tools within its newsroom, aiming to enhance editing, summarization, and content generation processes while maintaining a commitment to human oversight in journalism.

• The Times has introduced several AI tools, including Echo for summarizing articles and internal communications, alongside other platforms like GitHub Copilot, Google Vertex AI, and OpenAI’s non-ChatGPT API.

• Staff will undergo training in AI usage, with new editorial guidelines outlining appropriate applications such as editing suggestions, generating promotional copy, and creating news quizzes or FAQs.

• Restrictions on AI Use:

  - AI cannot draft or significantly alter articles.

  - No circumvention of paywalls or inclusion of copyrighted material from third parties is permitted.

  - AI-generated content must be explicitly labeled before publication.

• The Times reiterated that all journalism must be created and verified by professional journalists, affirming that the generative AI tools will only support, not substitute, human editorial efforts. This commitment was reinforced in a memo stressing that any AI use must begin with factual information vetted by reporters and be subject to editorial review.

⚒️ TOOL SNAPSHOTS

Futuristic tools within AI, no-code, and productivity

• 🛠️ Webdraw Beta - Create custom AI apps, no coding required. Free option available.

• 🧠 Tanka - Your team's AI messenger and memory bank. Free to use.

• 🎧 Midlife Engineering - Enhances mood and focus through ambient soundscapes. Free to use.

• ☁️ Browser Use Cloud - Automates social media and web tasks effortlessly. Payment required.

• 🤖 Base Chat - AI-driven, instant knowledge right from your company resources. Payment required.

📰 TOP NEWS

News on AI, no-code, automation, and productivity

🛠️ Fiverr Introduces AI Tools to Help Freelancers Monetize Their Work

Fiverr is launching AI-powered tools that let freelancers train models on their work, automate tasks, and sell AI-generated content, with a $25/month subscription for access.

📱 Humane Sells to HP and Shuts Down AI Pin

Humane is selling its assets to HP for $116 million and will discontinue the AI Pin, with cloud-based features shutting down on February 28, 2025.

🧠 Former OpenAI CTO Launches AI Research Lab

Mira Murati has founded Thinking Machines Lab, an AI research and product company focused on adaptable AI systems, open science, and multimodal capabilities.

🌍 Mistral Saba Brings AI Closer to Regional Languages

Mistral has introduced Saba, a 24B parameter AI model designed for Arabic and Indian-origin languages, offering faster, more accurate, and locally deployable solutions for businesses in the Middle East and South Asia.

ℹ️ ABOUT US

The Intelligent Worker helps you to be more productive at work with AI, automation, no-code, and other technologies.

We like real, practical, and tangible use-cases and hate hand-wavy, theoretical, and abstract concepts that don’t drive real-world outcomes.

Our mission is to empower individuals, boost their productivity, and future-proof their careers.

We read all your comments - please provide your feedback!