Hello, AI Enthusiasts!

Welcome to FavTutor’s Monday AI Recap! We’ve gathered all the latest and important AI developments from the past week in one place, just for you.

In Today’s Newsletter 😀 

• 60% of GPT 3.5’s Outputs are Plagiarized

• Stability AI’s latest Image Generator Model

• Adobe’s New AI Assistant Lets You Summarize PDFs

• 5 new AI Tools & 4 New AI Jobs

and more!

CHATGPT
🤯60% of GPT 3.5’s Outputs are Plagiarized, Says Report

A startling 60% of the results from OpenAI's GPT-3.5 showed evidence of plagiarism, according to a Copyleaks analysis.

Insights for you:

• The study examined more than a thousand GPT-3.5 outputs, covering 26 academic and creative areas such as computer science, psychology, physics, law, and the humanities.

• It was discovered that 59.7% of all texts created by GPT-3.5 contained some level of plagiarism.

• Out of the outputs, 45.7% had precise text matches with 46.5% featured paraphrasing from previously published materials.

PRESENTED BY PINECONE

What you need for better GenAI applications

Learn more from Pinecone Research on how Retrieval Augmented Generation (RAG) using Pinecone serverless increases relevant answers from GPT-4 by 50%.

- The larger the search data, the more “faithful” the results.

- RAG with massive on-demand data outperforms GPT-4 (without RAG) by 50%, even with the data it was explicitly trained on.

- RAG, with a lot of data, ensures state of the art performance no matter the LLM you choose, encouraging the use of different LLMs (e.g., open-source or private LLMs).

STABILITY AI
📷️ Stable Diffusion 3 is here, Stability AI’s Latest Image Generator Model

Stability AI introduced the Generative AI world to its latest generation of image-generating AI model, Stable Diffusion 3. SD3 has some good fixes and better architecture.

Insights for you:

• Stability AI announced Stable Diffusion 3, its latest addition to the family of text-to-image models.

• Leverages Diffusion Transformers and Flow Matching as a part of its working model.

• Improves the ability to write text and provides superior image quality.

GROQ AI
🤖 Groq AI Outshines GPT in terms of the World’s Fastest AI

Groq AI has recently become the world’s fastest AI model, leaving several AI giants in the dust, one of them being ChatGPT. With Groq’s unmatched speed of almost 500T/s and incredibly low latency, it has taken the internet by storm.

Insights for you:

• Groq AI uses a Linear Processing Unit (LPU) instead of the traditional GPU model used by all traditional generative AI software until now.

• These LPUs are made of hundreds of parallel processors for streamlined processing.

• It paves the way for a better future for AI models by highly reducing efficiency and maintenance costs.

ADOBE
📖 Adobe’s New AI Assistant Lets You Summarize PDFs

Now you can talk to your PDF and save time for your presentation! Adobe introduces AI Assistant, a conversational AI chatbot for their Acrobat & Reader users. It was built on the same AI and ML models that were also used in Acrobat Liquid Mode.

Insights for you:

• Adobe launches its conversational AI Assistant chatbot in beta.

• Can perform various tasks from summarizing documents to generating citations for responses.

• Will soon be a part of their new add-on subscription plan.

GPT-4
👀 GPT-4 Surpasses Human Eye Specialists at Diagnostic Questions

When it comes to detecting and treating eye disorders, GPT-4 can match or even surpass human ophthalmologists, as shown by the study from the New York Eye and Ear Infirmary of Mount Sinai (NYEE).

Insights for you:

• They contrasted the AI and human professionals' answers to a series of queries and patient scenarios about retinal diseases and glaucoma.

• Evaluation of the GPT-4's replies revealed that, especially in the case of glaucoma, they were on par with or better than those of the human professionals.

• GPT-4 even matched human professionals for questions about the retina and case scenarios, proving its capacity to accurately diagnose and provide therapies for retinal disorders.

GOOGLE
😮 Google Releases Gemma, a Lightweight Open-Source AI Model

Just a few days after revealing the greatness of Gemini 1.5, Google announced the release of its new generation of open models with Gemma. Google claims that it can run on any developer’s laptop computer, mobile, and cloud. Gemma can be used to create chatbots, summarize, and generate content, which is very efficient for low-resource environments.

Insights for you:

• Google has revealed Gemma in two different sizes namely Gemma 2B and 7B.

• Similar model architecture and deployment as that of their flagship Gemini. Built extensively for research purposes.

• Programmed with a Responsible Generative AI Toolkit and made accessible across a wide range of platforms through Colab notebooks and Kaggle.

AI TOOLS
⭐️ Amazing AI Tools to Try This Week:

• TechTok – With TechTok get an overview in under 5 minutes from the trending repos on GitHub, Hacker News, engineering blogs from leading tech companies, Reddit, and more. (link)

• Vadoo AI – Search across videos with this AI-powered Widget. (link)

• Talently Mock Interview – This tool helps you prepare for your next interview for any role in the world and get instant feedback to improve your performance and become a top candidate. (link)

• TacticMatrix – Combine time management and prioritization of tasks seamlessly with this tool. (link)

• Dart – A project management tool that helps you automate your tasks. (link)

AI JOBS
😍 New Job Openings This Week:

• Scale AI - Computer Science Expert AI Training (link)

• Cybernetics Control Limited - CS03 Backend / ML Developer (link)

• ICF - Senior Data Scientist (link)

• CFRA - Data Scientist, Quantitative Research (link)