The rise of ChatGPT has raised public awareness of AI and fueled the debate about the ethics of AI. How should AI be used? What are the implications for society, not just business?
Remember, the bias inherent in AI applications is related not only to how the algorithm was built and by whom, but also to how the model itself was built.
After all, there have been very public examples of political and gender bias exhibited by AI platforms. OpenAI CEO Sam Altman admitted last month that ChatGPT has “bias flaws.” However, these biases and mistakes can have far-reaching implications when applied to areas such as insurance platforms and drug discovery, and erroneous decisions can have far-reaching repercussions.
MLOps (a mashup of “machine learning” and DevOps) is a set of practices aimed at reliably and efficiently deploying and maintaining machine learning models in production and monitoring their biases. Simply put, MLOps practices are used by data scientists, DevOps, and machine learning engineers to transition AI algorithms into everyday operational models. The idea here is to improve model automation while also paying attention to business and regulatory requirements regarding bias and other aspects of AI. Improved efficiency also has a positive impact on the environment.
Seldon is a UK startup dedicated to this rare world of development tools for optimizing machine learning models. Competitors include Arise, Fiddler ($45.2 million in funding), Dataiku ($846.8 million in funding), and DataRobot ($1 billion in funding).
Seldon’s cloud-agnostic machine learning deployment platform has secured a £7.1m Series A from AlbionVC and Cambridge Innovation Capital in 2020.
It has now raised $20 million in a Series B round led by new investor Bright Pixel (formerly Sonae IM). Also participating were existing investors AlbionVC, Cambridge Innovation Capital and Amadeus Capital Partners.
Founders Alex Housley (CEO) and Clive Cox (CTO) claim 400% year-over-year growth in Seldon’s open source framework since its Series A in November 2020. This is important. Because Seldon is an open source network, you can distribute your own solutions far and wide. More efficient and cost effective.
“Seldon differentiates itself by presenting a unique solution that can reduce the user burden of deploying and explaining ML models in any industry. Combined with our risk and compliance capabilities, this means faster time to value,” said Pedro Carreira, director of Bright Pixel, in a statement.
Current Seldon customers include PayPal, Johnson & Johnson, Audi, Experian and more.
Alex Housley, founder and CEO of Seldon, said in an interview. “We are already well established in open source distributions, and what we have just validated is a new concept of data-centric MLOps with tight integration between data streams and production environments. You can improve your AI model, but there are small improvements to it, or this is our approach, but you can squeeze out much more performance by improving your data quality production. We have been working with the University of Cambridge and it has been a huge success.”
According to Run:ai’s State of AI Infrastructure Survey, 2023, more than half of these models have not been put into production at 88% of companies. why? Projects stall and duplicate work occurs across business silos.
Seldon claims it improves team collaboration and reduces deployment time by an average of 84%. This could be important given that more regulation is coming to AI (via EU AI ACT, US EEOC, etc.). Seldon and his competitors are racing to allow companies to improve these AI models internally while remaining compliant with these regulations.
The company works closely with Neil Lawrence, the first DeepMind Professor of Machine Learning at the University of Cambridge.
