Lippincott’s Post

11x more AI models successfully made it into production in the last year. Data and AI is hard but many of our customers are realising the value from their use cases having developed a robust approach to MLOps and actually getting them into production! Based on recent customer engagements had a gut feeling this was the case but it’s great to see this analysis of our customer base and the stats behind it.

Interesting post to review for any Data Professionals that are responsible for building and putting ML models into production! Our data shows that organizations have become 3x more efficient at putting ML models into production in just 14 months. They're spending fewer resources on experimental models that never provide real-world value. We break down these trends and more in the State of Data + AI, which examines usage data from our 10,000+ global customers. https://dbricks.co/3KB4OIK Activate to view larger image,

This is a great chart which shows just how efficient it is to get your data science team set up on databricks. If you want to see more DS projects make it to production, the right platform is the key! At Tempered AI, we have experienced consultants that can help your team accelerate adopting MLOps on Databricks.

Our data shows that organizations have become 3x more efficient at putting ML models into production in just 14 months. They're spending fewer resources on experimental models that never provide real-world value. We break down these trends and more in the State of Data + AI, which examines usage data from our 10,000+ global customers. https://dbricks.co/3KB4OIK

This is interesting stats thinking back to when studies pinned 80% of ML projects going into the graveyard for lack of deployment. Nowadays MLOps is a word, treating models as software services is the norm, and executive support is present by default. While excited to see the continued growth of the AI/ML space in industry... Similarly dreading the trough of disillusionment that's bound to come with the Gen AI craze of late. It'll not help when the investment blows away and the prudent ML practitioners gets dragged down by the business-driven cycle. Disclaimer: Gen AI is very powerful and useful albeit niche. 2018 feels really recent but all i could give: https://lnkd.in/ek4f5Xtf

I love this chart for two reasons. First, it shows how serious Manufacturers and Retailers - both process driven industries - have become about embedding ML into their business. ML is mainstream. Second, even industries that are less process driven and more experimental such as Financial Services (evaluating companies or investment hypotheses) and Health and Life Sciences (drug discovery) have refined their approaches. The initial wave of Machine Learning is over and companies are extracting real value.

Great news on organizations quickly becoming more efficient with putting machine learning models into production: Based on the ratio of (the number of) logged ML experiments to actually deployed/registered models, the average efficiency increased by a factor of 3 in only 14 months! The data for this evaluation stems from MLflow (as open source MLOps platform). And looking at the results separately, the number of registered models even increased by a factor of 10 (!) within a single year. The number of companies registering at least one model was up by 210%! All in all, great data to show that the GenAI wave also accelerates ML production adoption in general! Thanks Databricks for the insightful report! #mlflow #ml #ai

A New Survey Of Generative AI Shows Lots Of Work To Do http://dlvr.it/Sxb2k3

Atlanta is leading the way! Our recent study reveals that 71% of knowledge workers across the US are using generative AI at work, but in Atlanta that number is already 78%. See more about how tech is driving the future of work in our #WorkTrendIndex here: https://aka.ms/WTI24