GPT 4 is coming, and it's coming with a Bang

And Anthropic continues to raise funds to compete with OpenAI

Happy Friday ! Next week is going to be another massive week in AI, let's see if we can pre-empt what's about to happen.

As always, if you enjoy reading our posts be sure to spread the word !

Here’s what we have lined up for you today -

  1. GPT 4 is coming , and it's coming with a bang

  2. Anthropic is raising fresh funding at $4.1 billion valuation

  3. The Weekend level up

Some juicy news coming out of Microsoft as Microsoft Germany candidly spoke about the imminent release of GPT 4 at an AI kickoff event the company organised. Sam Altman, OpenAI CEO previously cautioned that like GPT 3, GPT 4 will not be anywhere close to AGI capabilities. The Microsoft team did share what users can expect from the model.

  1. "We will introduce GPT-4 next week, there we will have multimodal models”. Microsoft has been teasing at multimodal models with the release of their Kosmos -1 model. The model that can understand images, text, images with text, OCR, image captioning, and visual QA.

  2. GPT 4 is not about replacing jobs, but more so making AI models really good at jobs where repetitive patterns exist. This means that “Traditional job descriptions are now changing and exciting new professions are emerging as a result of the enrichment with the new possibilities.” according Microsoft Germany CEO.

  3. The executives hinted at some of the upcoming capabilities built into GPT 4. Specifically being able to successfully translate language into images, music and video.

  4. Microsoft Germany highlighted how their AI models are helping customers save 500 working hours a day through a speech to text product that they built. A product that was created by a single developer using GPT’s capabilities.

  5. Microsoft Germany highlighted that the three main use cases where models are being deployed heavily are, answering questions on company knowledge that is only accessible to employees, AI-assisted document processing and semi-automation by processing spoken language.

  6. Confidence metrics are also being worked on as the model still hallucinates answers and needs to be continuously verified.

If GPT 4's step change is anything like its predecessors, we can expect another exponential curve of progress kicking off.

Anthropic has been making the news alongside OpenAI on being one of the handful of companies that are building foundation models for AGI. They have a ChatGPT competitor out in the market called Claude, although it’s only available to limited users at the moment.

They have taken a slightly more cautious approach to deployment and haven’t had a mass rollout of their technology. They seem to be taking an enterprise sales approach to onboard customers.

They have however been publishing relentlessly with a lot of focus on AI policy and safety in their publications. Early in February, Google had announced that they were investing $300million into the startup. Just a month later it’s been made public that Anthropic is raising another $300 million from Spark Capital valuing their company at $4.1 billion. 

With no clear business model yet (although we could guess it might similar to OpenAI) the massive tranches of capital raised by the likes of Anthropic show that training and scaling these models definitely ain’t cheap.

Around the industry -

  1. Visual ChatGPT: Talking, Drawing and Editing with Visual Foundation Models

  2. AI chip startup Mythic rises from the ashes with $13M, new CEO

  3. Discord updates its bot with ChatGPT-like features, rolls out AI-generated conversation summaries and more

  4. AI trained on YouTube and podcasts speaks with ums and ahs

  5. US Chamber of Commerce calls for AI regulation

  6. Nat Friedman’s (Former Github CEO) Open Playground that compares how foundation models are performing in real time.

Product Watch Fini -

Convert your knowledge base into an AI Chat in 2 minutes. Supercharge customer support at your company.

Your favourite AI Tutor returns !

Weekends are for levelling up. We've shared Andrej's (OpenAI cofounder, ex Tesla AI head) walk throughs on neural networks and building GPT. And we think this weekend is a good time go deeper into the mechanics of building Language Models directly from the man himself. After all, GPT 4 will be coming out any moment now.