⚽ Has OpenAI finally met its match?

🚨 OpenAI deletes key NYT case evidence

AITR banner

TOGETHER WITH ARTISAN

Welcome to AI Tool Report!

Thursday’s top story: A Chinese AI lab has previewed its new AI ‘reasoning’ model, which is good enough to rival OpenAI’s o1.  

📢 Want to unlock your true potential with AI? You need to read “IRREPLACEABLE: the Three Competencies of the Future”, by Pascal Bornet

🌤️ This Morning on AI Tool Report

  1. ⚽ Has OpenAI finally met its match?

  2. ⚡ How to automate lead generation with Artisan

  3. 🚨 OpenAI deletes key NYT case evidence

  4. 🤖 How to build custom chatbots with Chatnode 

  5. 🎯 How to create an effective ad campaign using ChatGPT

  6. 😍 How to make eye-catching presentations with AI

  7. 📉 Is NVIDIA set for huge downfall?

  8. ⬆️ ChatGPT-4o gets a major upgrade

Read Time: 5 minutes

FACT OF THE DAY

🤔 According to Superside, the creative industry has the highest rate of AI adoption, with 83% of creative professionals already using AI tools and 76% believing that AI will be essential to creative processes over the next 5 years.

STOCK MARKETS

AI stock tracker

💥 AI and tech stocks slid to the downside to start the day but recovered well, ending the session almost neutral. NVIDIA released its earnings report at market close, showcasing the power of the AI-boom. Data center revenue was highlighted, up 112% Y/Y and 17% Q/Q. The CFO expects Blackwell chip demand to exceed supply for several quarters. Learn more.

  — — — — — — —

ADVANCEMENTS

Has OpenAI finally met its match?

Our Report: Chinese AI lab—DeepSeek—has released a preview of a new AI model, DeepSeek-R1, which experts are saying is one of the first “reasoning” AI models capable of rivaling OpenAI’s o1 reasoning model.

🔑 Key Points:

  • Like o1, DeepSeek-R1 spends more time (up to 10 secs) evaluating and fact-checking its own responses before answering, emulating human-like thought processes, and reducing inaccurate or nonsensical answers.

  • It reportedly performs on par with o1 on the AIME benchmark (uses AI to evaluate performance) and MATH benchmark (tests ability to solve word problems), and both struggle with logic games like tic-tac-toe.

  • But, users have found weaknesses in DeepSeek-R1’s safeguards and content restrictions—it avoids politically sensitive topics—which are likely due to government pressures for AI models to “embody core socialist values.

🤔 Why you should care: This comes as recent reports suggest that AI models, from the likes of OpenAI, Google, and Anthropic, aren’t improving at the scale they used to—in part due to a lack of new and available training data—which has led to the exploration of new improvement techniques, like test-time compute (gives models additional processing time to complete tasks more effectively), which is underpinning both o1 and DeepSeek-R1.

   — — — — — —

TOGETHER WITH ARTISAN

Artisan image

Artisan unifies your outbound sales tools into one platform, featuring Ava — the AI Business Development Rep who manages it all.

With Artisan, you’ll benefit from:

  • Access to 400M+ high-quality B2B prospects, including Local Business Data & E-Commerce

  • Lead enrichment across 10+ data sources, including website visitor tracking

  • Advanced personalization via LinkedIn, Twitter, and web scraping

  • Automated outbound sequences across email and LinkedIn

  • Comprehensive email and LinkedIn deliverability management tools

   — — — — —

LEGAL

OpenAI deletes key NYT evidence

Our Report: Lawyers, working on the New York Times (NYT) lawsuit against OpenAI—for using copyrighted material to train AI models without permission—have revealed that OpenAI engineers have deleted key data that might have been relevant to the case.

🔑 Key Points:

  • OpenAI gave NYT lawyers access to its AI training data so they could search for copyrighted content within the datasets, and since Nov 1st, hired experts have spent over 150 hours doing just that.

  • However, on Nov 14th, OpenAI engineers ‘accidentally’ deleted the entire repository of search data (which could have contained potential evidence), which was stored on one of OpenAI’s virtual machines.

  • Although OpenAI recovered most of the data, what remains is unusable as it lacks the contextual information necessary for proving copyright infringement, meaning 150 hours worth of work will need to be redone.

🤔 Why you should care: NYT lawyers don’t believe this was done intentionally to sabotage the case, but they will incur more costs as a result, while OpenAI stands firm on their stance that using publicly available data (including news articles from the NYT) to train their models is fair, although they have recently inked several lucrative publishing deals, permitting them to use their content for training purposes.

Does OpenAI have a point: Should they be able to use published work to train AI models?

Login or Subscribe to participate in polls.

    — — — —

TOGETHER WITH CHATNODE

Still drowning in customer questions?

With Chatnode, it's now even easier to build custom, advanced AI chatbots that deeply understand your business, handle inquiries 24/7, and drive more sales.

Here’s why Chatnode is different:

🔍 Reliable Responses: RAG technology ensures consistent and accurate answers.

Easy Setup: Create and launch your chatbot quickly, no coding required.

🔒 Enterprise-Grade Security: Top-tier security and compliance for your data.

🔌 Connects with Your Software: Zendesk, Slack, Google Drive, Notion, Zapier, SharePoint, Dropbox, Onedrive, Make, and more.

💬 Live Agent Handoff: Easily transfers chatbot conversations to human agents.

🤖 Model Agnostic AI: Choose your LLM: Claude, Gemini, ChatGPT, and Perplexity.

PROMPT ENGINEERING

     — — —

Thursday’s Prompt: How to create an effective ad campaign using ChatGPT

Type this prompt into ChatGPT:

Write an ad campaign outline using the 'PAS' formula to identify the problem faced by [ideal customer persona], agitate that problem to make it more pressing, and then present our [product/service] as the solution

Copy Frameworks

Results: After typing this prompt, you will get an ad campaign outline based on the PAS framework which identifies the key problem your key customer has, exacerbates this issue, and then presents your product/service as the ideal solution.

P.S. Use the Prompt Engineer GPT by AI Tool report to 10x your prompts.

ACTIONABLE INSIGHTS

       

AI Report image
  • Creating eye-catching slides that capture attention and set the tone of your presentation is difficult without the help of AI tools, like the ones below.

  • Use Gamma to create visually engaging, professional slide layouts, based on your topic, without any design expertise needed.

  • Beautiful AI has templates with built-in design rules that adjust spacing, suggest colors, recommend images, and keep branding consistent.

  • Using AI tools like these sets a consistent and professional tone for your presentation, allowing you to focus on the key messages.

BREAKING NEWS

    

AI HARDWARE

  • Although NVIDIA is the most valuable company in the world, investors are concerned about its continued rapid growth as AI companies find new ways (namely, test-time compute, as above) to improve their AI models.

  • To reassure investors, NVIDIA CEO, Jensen Huang, called test-time compute an “exciting development” that would play a broader role in NVIDIA’s strategy, indicating it was well-positioned for the change. 

  • But giving AI models more time to “think” means more AI inference chips, not training chips—NVIDIA’s bread and butter—and other chip manufacturers are busy developing lightning-fast inference chips, as we speak.

    — — —

CHATBOTS

  • Following its initial launch in May, OpenAI has released a major upgrade for ChatGPT-4o that has reportedly brought improvements in two different areas: Creative writing and file handling.

  • The updated version now generates more natural, engaging, and relevant writing that’s more tailored, fluid, and human-like, making it more valuable for content creators looking to create high-quality text.

  • GPT-4o can already work with uploaded files—like PDFs and text files—but it can now understand complex material better, allowing it to deliver more thorough and detailed insights and responses. 

🕊️

📖 

📺

We read your emails, comments, and poll replies daily.

Hit reply and tell us what you want more of!

Got a friend who needs to learn more about AI? Sign them up to the AI Tool Report, here.

Until next time, Martin & Liam.

P.S. Don’t forget, you can unsubscribe if you don’t want us to land in your inbox anymore.

What did you think of this edition?

Login or Subscribe to participate in polls.