🗓 Next week in TheSequence Edge:

Edge #263: Our series about ML interpretability dives into counterfactual explanations, Google StylEx method that can generate explanations from image classifiers and Microsoft’s open source implementation of the DiCE method.

Edge #264: Deep dive into Google’s newest generative AI model: Muse.

📝 Editorial 

The strategic relationship between Microsoft and OpenAI has been capturing headlines in light of a potential $10 billion investment in OpenAI’s new tender offer. Years ago, Microsoft has the foresight to invest $1 billion in OpenAI in 2019 and proceed to integrated its API across different product units such as Azure or GitHub. Part of the benefits of that investment materialized last week when Microsoft announced the general availability of the Azure OpenAI Service. The Azure-native service, which has been in Beta for quite some time, features API access to OpenAI models such as GPT-3.5, DALL-E or Codex. Together with the announcement, Microsoft disclosed that very soon the Azure OpenAI Service will include programmatic access to the uber popular ChatGPT model.

The alliance between Microsoft and OpenAI is one of the most fascinating strategic developments in the AI world. Obviously, the partnership provides Microsoft with first-level access to what many consider the most innovative generative AI stack in the market. A more subtle and often overlooked aspect of this partnership is how much Microsoft can contribute to OpenAI beyond capital. Embedding OpenAI capabilities in platforms like Office365, Windows, Bing, Power Platform will provide, arguably, the richest distribution ever achieved by generative AI technologies across both consumer and enterprise segments. Also, Microsoft Research is one of the most sophisticated AI labs in the world and can also contribute to advance OpenAI’s agenda.

The partnership with OpenAI is making Microsoft one of the most exciting companies in tech and can provide the ideal launchpad for the next wave of generative AI applications.

🔎 ML Research

Google Research Outlook

Google Research published an amazing summary of their research efforts in 2022 and their vision for 2023 —> Read more.

Transformers for Open Vocabulary Object Detection

Carnegie Mellon University published a paper describing an object-detection method in images based on complex descriptions —> Read more.

Human-Aligned AI Research

Microsoft Research published a summary of their recent research efforts in responsible AI —> Read more.

Teaching Speech Recognizers New Words

Amazon Science published a paper proposing a technique to teach speech models to recognize new entities without retraining —> Read more.

🤖 Cool AI Tech Releases

Azure OpenAI Service GA

In the biggest AI news of the week, Microsoft announced the general availability of its Azure OpenAI Service that includes models like GPT-3.5, Codex and DALL-E —> Read more.

Write

After raising $100 million, NLP startup DeepL announced the release of Write, a new AI-based writing assistant —> Read more.

PaddlePaddle and HuggingFace

Baidu’s incubated PaddlePaddle ML framework announce the initial integration with the HuggingFace Hub —> Read more.

🛠 Real World ML 

Complete Looks at Walmart

Walmart Global Tech discusses the ML architecture powering the interface to develop personalized stylized outfits —> Read more.  

Reference Architecture for Semantic Segmentation

The TensorFlow team published a reference architecture for semantic segmentation pipelines using TFX, Hugging Face and Google Cloud —. Read more.

💸 Money in AI

• AI startup Impel raised $104 million to expand their platform for customer engagements in auto dealers.

• AL sales automation startup CloseFactor announced a $15.2 million series A to deliver insights across different sales data sources.

• Rider raised $73 million to continue building autonomous yard trucks.

• Generative AI Scenario raised $6 million to let artists build highly tailored game assets.