Back to blog home

AI and MLOps Roundup: September 2023

Microsoft is ready for big tech's war for AI dominance. But will hallucination snowballing ruin LLMs? And what makes an AI engineer different than others? Check out our roundup of the top AI and MLOps articles for September 2023!

Microsoft’s Satya Nadella is winning Big Tech’s AI war

When Satya Nadella became CEO of Microsoft, its market cap was a stagnant $300 billion; today it has surged past $2.5 trillion. Here's how he plans to become the leader in AI: https://www.fastcompany.com/90931084/satya-nadella-microsoft-ai-frontrunner

Microsoft CEO Satya Nadella

The Rise of the AI Engineer

The "AI Engineer" is an emerging subdiscipline separate from ML engineers, data engineers, or other software engineering roles. Here's what makes it different and why it will be the most in-demand skill: https://www.latent.space/p/ai-engineer

The API line is permeable - AI Engineers can go left to tweak/host models and Research Engineers go right to build atop APIs too! But their relative strengths and “home base” is clear.

Four Ways that Enterprises Deploy LLMs

With the rapid pace of LLMOps, so far enterprises have deployed LLMs in four ways. Learn the pros and cons of each, and which may be the right fit for your org: https://www.fiddler.ai/blog/four-ways-that-enterprises-deploy-llms

Four approaches to LLMs in production

How Language Model Hallucinations Can Snowball

A major risk of LLMs is their tendency to hallucinate, which can result in further false claims and eventually hallucination snowballing. Researchers explored why this happens and how to fix it: https://arxiv.org/pdf/2305.13534.pdf

Model hallucinations may be induced by an earlier hallucination for consistency

Chronon — A Declarative Feature Engineering Framework

Airbnb went from months to days to generate new sets of features for their models. Learn how they're able to continuously turn raw data into features for both training and serving: https://medium.com/airbnb-engineering/chronon-a-declarative-feature-engineering-framework-b7b8ce796e04

Chronon at a high level

Testing Language Models (and Prompts) Like We Test Software

Testing LLMs like software is a particularly challenging emerging area of focus. Check out this great guide on ways to get started: https://towardsdatascience.com/testing-large-language-models-like-we-test-software-92745d28a359

LLM testing is hard

Stay up to date on everything AI and MLOps by signing up for our newsletter below.