Salesforce recently filed for a patent covering “automatic machine learning model generation”. This system aims to automate the creation of AI models, making it simpler for users to develop machine learning abilities. The system analyzes user data and its associated metadata, generates a prediction field from this dataset, selects and trains a machine learning model accordingly, and delivers ready-to-use predictive models back to users. Salesforce strives to simplify machine learning development for users without advanced statistical training, with manual data entry or incomplete/incorrect input potentially leading to inaccuracies or failed predictions; their system seeks to mitigate such issues by automating every aspect of this process.
Automatic machine learning has long been of interest to developers. Oracle and JPMorgan Chase have explored similar concepts through patent filings related to chatbot-generating chatbots and no-code AI capabilities, while Salesforce is working on automatic machine learning capabilities as part of Einstein, its AI activities umbrella. While Salesforce may not boast as many AI clout as Google, Microsoft, or Meta do, the company has actively been creating AI tools for analytics, customer data processing, automation, and data collection within Einstein; with this patent filing they hope to provide a seamless user interface for machine learning development development.
Experts argue that some of Salesforce’s patent capabilities may already be outdated. Large language models like GPT (Generative Pre-trained Transformer) can achieve similar results by breaking complex problems down into manageable components and providing solutions. Furthermore, these models do not require as much manual intervention – which provides results to use by decision-makers more quickly than before. Salesforce is working on an enterprise-focused language model called Einstein under its AI umbrella to achieve similar goals.
Salesforce’s patent filing for automatic machine learning model generation shows their commitment to streamlining the machine learning development process, offering users without specialist statistical training an easier way to leverage AI capabilities. Large language models may offer other alternatives with less manual intervention; as Salesforce continues expanding their AI tools and capabilities, it will be interesting to observe if their patent application aligns with their overall AI strategy.