What should be the consequences of the increasing use of AI? What ethical considerations should be taken into account when implementing AI technologies in our society?
We must be very cautious when working with AI applications, as the inherent prejudices that come from how the algorithm and model are built can lead to undesirable outcomes in this new AI environment.
Recently, instances of political and gender inequity by AI systems have been well-known. Sam Altman, CEO of OpenAI, confessed only last month that ChatGPT had “issues with bias”. However, if that bias was to be applied to important fields like insurance and drug development, the consequences of making wrong decisions could be devastating.
MLOps, a term combining “Machine Learning” and DevOps, is a strategy for deploying, maintaining, and monitoring machine learning models in production. It is used by Data Scientists, DevOps, and Machine Learning engineers to move AI algorithms into everyday production models. This process helps to automate the model while simultaneously monitoring the business, regulatory, and bias requirements of AI. Enhancing productivity has a beneficial effect on the environment.
Seldon, a U.K. based business focused on creating specialized tools for enhancing machine learning models, faces competition from Arise, Fiddler (which has raised $45.2 million), Dataiku (which has raised $846.8 million) and DataRobot (which has raised $1 billion).
In 2020, Seldon’s platform for deploying machine learning regardless of the cloud provider received a $7.1 million Series A investment from AlbionVC and Cambridge Innovation Capital.
A new investor, Bright Pixel (formerly Sonae IM), has led a $20M Series B funding round which was also joined by existing investors AlbionVC, Cambridge Innovation Capital, and Amadeus Capital Partners.
Alex Housley (CEO) and Clive Cox (CTO) boast that Seldon’s open source frameworks have increased fourfold year-on-year since its series A in November 2020. This is noteworthy as the open source network makes it possible to spread their exclusive solutions faster and more economically.
Bright Pixel’s Director, Pedro Carreira, has noted that Seldon has created a special solution that eliminates the difficulty of utilizing and understanding ML models in any industry. This solution enables customers to become more productive, quickly achieve results, and stay compliant with economic regulations.
Seldon’s clientele consist of well-known names such as PayPal, Johnson & Johnson, Audi, and Experian, to name a few.
Alex Housley, Seldon’s Founder and CEO, said in an interview that AI is everywhere and Seldon is in a good spot. He also mentioned that their open source distribution is already very successful and they just recently validated a novel idea of MLOps that is closely linked to data streams and production. To simplify, you can slightly upgrade an AI model with its algorithm, but the impact is minor. Rather than taking a different approach, we have been able to get better performance by enhancing the production of data quality. We have had excellent results while working with Cambridge University on this.
The survey done by Run:ai found that 88% of companies have AI models that never reach the point of production. The reason for this is usually because the projects grind to a halt or there are too many people working on the same thing in different departments.
Seldon claims its services can help teams work together more efficiently, resulting in an average decrease of 84% in deployment time. This is especially relevant with the numerous regulations on AI (e.g. the EU AI ACT and US EEOC). Seldon and its competitors are competing to ensure businesses stay compliant with these regulations while still improving their AI models.
The company has worked in conjunction with Neil Lawrence, who is the first DeepMind Professor of Machine Learning at the University of Cambridge.
