Data and AI newsletter: March 2023
The release of The GPT-4 marks a significant advancement in scaling deep learning and the large multimodal model.
On March 14, the Anthropic introduced Claude - a next-generation AI assistant based on Anthropic's research to train helpful, honest, and harmless AI systems. It was created by ex-OpenAI researchers and is meant to be a competitor to the Chat GPT. Moreover, MIT's first-ever Data-Centric AI course is free and includes algorithms to find and fix common problems in ML data and construct better datasets.
It's time for our monthly dose of data and AI news. So, read all the interesting information and resources in our data and AI newsletter.
NEWS AND INSIGHTS
The GPT-4 model, also known as Generating Pre-trained Transformer 4, is the fourth version of the OpenAI-created GPT family of large language models. It is the GPT-3 model's successor, and it drives the popular AI chatbot ChatGPT. Like its predecessor, GPT-3, GPT-4 is intended to produce text that looks like it was written by a human, carry out tasks like language translation and summarization, and even generate creative writing like fiction, poetry, and song lyrics.
"GPT-4 can generate text and accept image and text inputs — an improvement over GPT-3.5, its predecessor, which only accepted text — and performs at "human level" on various professional and academic benchmarks. For example, GPT-4 passes a simulated bar exam with a score around the top 10% of test takers; in contrast, GPT-3.5's score was around the bottom 10%."
In all the noise about Google's new generative AI features or the mighty GPT-4 launch, you may have missed the launch of Claude, a new chatbot by Anthropic meant to rival ChatGPT.
- created by ex-OpenAI researchers
- backed with $300M by Google
- available to end users and developers via API
- less likely to produce harmful outputs, easier to converse with, and more steerable
AI may assist in a variety of ways to assure the best possible driving experience, not just in the development of self-driving vehicles. The businesses collaborate to create intelligent automobiles, which automatically recognise you and customise their settings to suit your tastes. The vehicles can also simply make up a video chat for you if necessary and explain in plain terms what is wrong with them.
▶️ The first-ever full course on Data-Centric AI by MIT is free and available online
"What does Data-Centric AI (DCAI) mean? An emerging science that studies techniques to improve datasets, which is often the best way to improve performance in practical ML applications. While good data scientists have long practiced this manually via ad hoc trial/error and intuition, DCAI considers the improvement of data as a systematic engineering discipline.
This is the first-ever course on DCAI. This class covers algorithms to find and fix common issues in ML data and to construct better datasets, concentrating on data used in supervised learning tasks like classification. All material taught in this course is highly practical and focused on impactful aspects of real-world ML applications, rather than mathematical details of how particular models work. You can take this course to learn practical techniques not covered in most ML classes, which will help mitigate the "garbage in, garbage out" problem that plagues many real-world ML applications."
Built by OpenAI on the Slack platform, the app integrates ChatGPT's powerful AI technology to deliver instant conversation summaries, research tools, and writing assistance.
INTERESTING AI STARTUPS AROUND
Hungarian text-to-video platform Colossyan gets $5M in its Series A round to build up its R&D/engineering teams and expand to London and New York. The round was led by LAUNCHub Ventures and joined by Emerge Education, Day One Capital, Oktogon Ventures, and APX.
Mily Tech collects, connects, and enriches delivery data with spatial context, making it easy to gain insights and quickly identify possibilities for improvement. Serbian analytics platform obtains €1M pre-seed round.
💡 Resources on Multimodal ML
On March 1st, we were honoured to host the PyData Sofia Meetup where one of the top data scientists in the country Georgi Nalbantov, PhD gave an exceptional presentation on the bias-variance trade-off conception. Read our recap article, including his video from the presentation.
"As generative AI models grow larger and more powerful, some scientists advocate for leaner, more energy-efficient systems." - learn more in Nature's article.
The Institute for Computer Science, Artificial Intelligence and Technology (INSAIT) is launching a series of lectures on the most sizzling topics, such as Neurosymbolic AI, Generative AI, and Geometric Deep Learning, directly from the technology leaders, researchers and entrepreneurs who create them. The upcoming talk will be given by prof. Armando Solar-Lezama in Sofia, on 03.04.2023. To attend it, register at techseries.insait.ai/.
Thanks for reading our monthly digest! If you enjoy it, we'd love your help spreading the word! Share it with friends and colleagues who might benefit from it.
The topics from the previous newsletters, you can find at: