This AI newsletter is all you need #33
Last Updated on February 13, 2023 by Editorial Team
What happened this week in AI by Louis
ChatGPT and generative AI are still hot topics this week. ChatGPT reportedly reached 100 million monthly active users in January, making it one of the fastest-growing apps in history. This success is leading to increased pressure on incumbent companies to integrate the latest LLM technology into their products. Open AI has launched several integrations with Microsoft as well as details of its subscription plan for ChatGPT. Google has also announced the upcoming release of its new ChatGPT competitor called “Bard” which is built on its LaMDA model.
After its partnership with OpenAI, Microsoft is preparing to launch a version of its Bing search engine using ChatGPT, aiming to become more competitive with Google. ChatGPT has become more accessible through integration into Microsoft’s release of Teams Premium. OpenAI is also planning to launch a mobile GPT app and test a video-generating AI. The rapid adoption of this technology also raises new questions and Microsoft, GitHub, and OpenAI are facing a proposed class action complaint that accuses them of scraping licensed code to build GitHub’s AI-powered Copilot tool. The companies have asked the court to dismiss the complaint.
OpenAI has launched a pilot subscription called ChatGPT Plus, to monetize its viral success. The service offers faster response times, access to ChatGPT during peak times, and priority access to new features and improvements. It starts at $20 per month. The early rumors were that the ChatGPT Plus feature would cost $42 per month, and many people believed it was too expensive. However, now that it’s priced at $20 per month, do you think it’s worth subscribing to?
This issue is brought to you thanks to Qdrant:
Qdrant open-source vector search engine launches managed cloud platform
Qdrant — robust vector similarity search engine with advanced filtering support. It is written in Rust, which ensures stability and high performance proved by benchmarks.
The managed cloud platform is now fully available for business use, allowing companies of any size to benefit from Qdrant’s cutting-edge features without handling its deployment and maintenance.
Qdrant cloud platform can be accessed through the website.
ChatGPT has been a topic of discussion since its launch, with opinions divided between its potential benefits and perceived threats. While some experts see ChatGPT as a tool that could greatly enhance AI-human partnerships, many people are unsure of how to react to it. This article asked the robot itself, what impact the ChatGPT could have on the engineering profession.
Small companies are driving AI to the masses, prompting Big Tech to respond. Three months before ChatGPT’s launch in November, Meta, Facebook’s parent company, introduced a similar chatbot, Blenderbot. However, Blenderbot failed to create the same excitement as ChatGPT. According to Yann LeCun, Chief Artificial Intelligence Scientist at Meta, the reason it was boring was that it was made safe.
In this Twitter thread, @alexandr_wang shared a list of the most interesting projects, including the winning projects created by the ~300 hackers who attended Scale AI’s Generative AI hackathon last week.
TheGP is seeking to connect with talented hackers and quick prototypers who are actively contributing in the fields of engineering, product, or design as part of the AI Literacy project. They aim to share and publish insights gathered from various builders and provide further learning opportunities by delving into deeper discussions on topics such as accessibility & application of language models, toolkits they use, and enhancing productivity & creativity with LLMs.
2022 was the year when generative AI left the lab and gained widespread recognition. Gary Marcus reflects on what has and hasn’t changed in recent years and provides a series of examples demonstrating the ease with which generative AI can produce nonsensical results and its superficial understanding of reality.
Three 5-minute reads/videos to keep you learning
The tutorial was initially hosted via a live stream on our Learn AI Discord. It teaches you how to build Buster, a chatbot that answers questions related to the Hugging Face transformers library while referencing its sources. It outlines the three critical components required to make Buster function: gathering data from the documentation, constructing a document retrieval system & ranking the most relevant sources, and generating text based on the user’s question and providing the answer.
This GitHub guide includes a collection of recent papers, educational resources, datasets, and tools relevant to prompt engineering. It also features a compilation of blog posts and books for further learning.
The article discusses how ChatGPT can enhance strategic thinking and decision-making capabilities, such as anticipating and planning for the future, thinking critically and creatively about complex problems, and making effective decisions in uncertain situations.
Nishith Agarwal, the Head of Data and ML Platforms at Lyra Health and the creator of Apache Hudi, draws on his experiences at both Uber and Lyra Health to present five considerations that impact the decision to build or buy the data warehouse, data lake, and data lakehouse layers of a data stack. These considerations include cost, complexity, expertise, time to value, and competitive advantage.
In this Twitter thread, Cohere presents a step-by-step guide on how to construct a basic semantic search engine. The steps involved include obtaining a list of texts to search, embedding the archive of questions, converting the embeddings into indexes, and others.
Enjoy these papers and news summaries? Get a daily recap in your inbox!
The Learn AI Together Community section!
Upcoming Community Events
The Learn AI Together Discord community hosts weekly AI seminars to help the community learn from industry experts, ask questions, and get a deeper insight into the latest research in AI. Join us for free, interactive video sessions hosted live on Discord weekly by attending our upcoming events.
This week’s session in the (free) nine-part Neural Networks Architectures series will be led by Pablo Duboue (DrDub) and focuses on Recurrent Networks. During this session, he will explore topics such as training RNNs by unrolling, internal memory access and update in LSTMs and GRUs, as well as Encoder/Decoder and attention in Encoder/Decoder systems. Find the link to the seminar here or add it to your calendar here.
Date & Time: 7th February, 11 pm EST
If you missed the first part of the series, find last week’s event recordings here.
Learn AI Together’s weekly reading group offers informative presentations and discussions on the latest developments in AI. It is a great (free) event to learn, ask questions, and interact with community members. Join the upcoming reading group discussion here.
Date & Time: 11th February, 10pm EST
Please reach out and pitch us your idea if you’d like to present any AI-related content to our community, whether it is a class, tutorial, paper, reading group, panel discussion or competition. We would love to host it live on our 38,000-member Learn AI Discord community. The content will be recorded, 100% owned by you, and available for you to distribute on your own Youtube channel or website afterward.
Meme of the week!
Meme shared by Louis B#1408
Featured Community post from the Discord
jUMAD1#0227 recently updated a data science notebook on linear regression for WW2 temperatures. The notebook was part of the 2019 Machine Learning Scholarship by Indonesia’s Ministry of Communication and Informatics. The update includes improved structure and code, a review of linear regression assumptions, and evaluation metrics. Find the notebook here and support a fellow community member! Join the conversation and share your feedback here.
AI poll of the week!
TAI Curated section
Article of the week
The first AI system for discovering novel, efficient, and mathematically proven algorithms for fundamental tasks such as matrix multiplication. This system sheds light on a 50-year-old open question in mathematics regarding the fastest method for multiplying matrices. This paper, released in 2022, is a fundamental breakthrough in machine learning and attempts to answer a translational research question with domain-agnostic applications and implications. The author has explained the paper in the simplest possible manner.
Our must-read articles
If you are interested in publishing with Towards AI, check our guidelines and sign up. We will publish your work to our network if it meets our editorial policies and standards.
If you are preparing your next machine learning interview, don’t hesitate to check out our leading interview preparation website, confetti!