• AI Breakfast
  • Posts
  • How to Use ChatGPT to Navigate Around Paywalls

How to Use ChatGPT to Navigate Around Paywalls

Step-by-step instructions inside

Author

AI Breakfast
June 26, 2023

Good morning. It’s Monday, June 26th.

Did you know: Window’s 98 was released 25 years ago today?

In today’s email:

  • How to Use ChatGPT Plus to Navigate Around Paywalls

  • Meta AI Powers Adult Chatbots

  • Andreessen Advocates AI Use for Increased Youth Productivity

  • NIST Initiates Public Working Group for Generative AI

  • DeepMind's Upcoming Algorithm Targets Surpassing ChatGPT

  • ChatGLM from China Excels over GPT-4?

  • 5 New AI Tools

  • Latest AI Research Papers

You read. We listen. Let us know what you think of this edition by replying to this email, or DM us on Twitter.

Today’s edition is brought to you by:

Attending Ai4 this year? We will be!

Join the world's premier AI for business community at Ai4, uniting the global AI ecosystem through events, media, and resources​.

Attend multi-day, networking-enabled conferences featuring the world’s most innovative speakers, attendees, and AI companies​.

Today’s trending AI news stories

How ChatGPT Can Navigate Around Paywalls

A recent discovery shows that ChatGPT Plus users can use the GPT-4 Browsing feature to access paywalled content on many online blogs and publications.

Here’s how to do it:

  1. Sign up for ChatGPT Plus ($20/mo)

  2. Enable Browsing Mode with GPT-4

  1. Use the following prompt: print the content of this article:

  1. Paste the URL after the prompt

  2. GPT-4 Browsing will display the text content of the article:

We signed up for Fortune’s subscription to confirm that the result from GPT-4 wasn’t a hallucination based on the article title, and the response from ChatGPT was, in fact, an exact reprinting of the paid article.

This capability may be patched at some point, as it violates the terms and services of many online subscription-only publications, but for now it appears that ChatGPT Plus users have free access of the paywalled publishing.

Try it out, and let us know what your results are by replying to this email, and we’ll do a follow-up later this week.

Quick News

Meta’s new AI is being used to create sex chatbots: Meta has developed a new AI integration that enables individuals to create their own chatbots. As expected, these models are being used by some to create explicit adult-oriented content.

Marc Andreessen says we’re in a ‘freeze-frame moment’ with A.I.—and has advice for young people: The co-founder of venture capital firm Andreessen Horowitz argues that artificial intelligence tools like ChatGPT and GPT-4 have the potential to dramatically increase productivity, create new industries and jobs, and usher in a new era of prosperity, encouraging young people to utilize these tools to stand out and achieve a hyper-productivity curve early on

NIST Launches Generative AI Public Working Group: The National Institute of Standards and Technology (NIST) is creating a Public Working Group on Generative AI to provide guidance, support testing and evaluation, and explore productive uses for generative AI technologies in areas such as health and climate change, based on the recently released NIST AI Risk Management Framework.

DeepMind’s CEO Says Its Next Algorithm Will Eclipse ChatGPT: DeepMind, the AI lab that made history with its AlphaGo program, is developing a new AI system called Gemini that combines the language capabilities of large models like GPT-4 with the problem-solving and planning techniques used in AlphaGo, with the aim of surpassing the capabilities of OpenAI's ChatGPT.

China's ChatGLM just beat GPT-4: This Twitter user acknowledges that the Chinese model ChatGLM has likely surpassed GPT-4 in specific Chinese benchmarks, but not in all areas, also noting that the top-performing version of ChatGLM is a closed model with 130B parameters, and expressing admiration for the rapid evolution and organization of Chinese teams working on AI models.

🎧 Did you know AI Breakfast has a podcast read by a human? Join AI Breakfast team member Luke (an actual AI researcher!) as he breaks down the week’s AI news, tools, and research: Listen here

5 new AI-powered tools from around the web

NExTNet is a cloud-based platform for scientists to uncover connections, generate hypotheses, and explore biomedical data. Its intuitive GUI enables the exploration of publications, drugs, genes, and pathways. Semantic search, no-code data access, and live collaboration enhance communication and decision-making efficiency.

Hourone is an AI-powered video generator that can produce professional videos. Leveraging text-to-video technology, AI script assistants, customizable templates, virtual human presenters, multi-lingual support, collaboration tools, and brand kits, it crafts high-quality videos in under 10 minutes.

Gan.ai empowers businesses to create personalized videos at scale, driving hyper-engagement, hyper-conversion, and hyper-awareness. With support for diverse industries like real estate, healthcare, B2B, e-commerce, and more, its features include user resurrection, increased LTVm conversion, customer engagement, personalized ads, and event-driven video generation.

StakWallas AI is an AI-powered freelancing marketplace that creates customized no-code business solutions. It utilizes AI to match you with the perfect freelancer and provides an instant price quote.

ImageCreator is a free AI-powered Photoshop plugin that empowers artists to easily create high-quality images. With features like TXT2IMG, IMG2IMG, Fill, and ControlNet, you can generate intricate images using AI models and prompts.

arXiv is a free online library where scientists share their research papers before they are published. Here are the top AI papers for today.

This research introduces the Retrieval-Pretrained Transformer (RPT), a language model that combines retrieval and generation capabilities. Unlike previous approaches, RPT trains the retrieval component together with the language model from scratch. It improves the quality of retrieval and the overall performance of the model across different language tasks. RPT achieves this by incorporating a self-retrieval, fusion of retrieved information, and training the retriever to retrieve relevant text chunks for better prediction.

DreamEditor is a framework that allows users to edit 3D scenes using text prompts. It represents scenes as mesh-based neural fields, enabling localized editing while preserving irrelevant areas. By leveraging a pretrained text-to-image diffusion model, DreamEditor identifies the regions to be edited based on text semantics and optimizes the geometry and texture alignment. Extensive experiments show that DreamEditor achieves precise and high-quality editing, surpassing previous methods in both quantitative and qualitative evaluations.

This paper discusses the importance of natural language feedback in improving user experience and proposes a framework for leveraging NL feedback at the system level. The existing approach focuses on refining specific examples, while this framework aims to incorporate feedback to make system-level design decisions and improve models. The paper presents two case studies on search query generation and dialog response generation, demonstrating the effectiveness of using system-level feedback. The study also highlights the benefits of combining system-level and instance-level feedback and emphasizes the importance of human feedback in building robust systems.

Thank you for reading today’s edition.

Your feedback is valuable.


Respond to this email and tell us how you think we could add more value to this newsletter.