Th use of generative AI solutions and AI-solutions lead to efficiency gains. Corporates such as TCS, HCL, Infosys and LTIMindtree seek to explore the building of their own version of generative AI consortium-based solutions. So far the regulatory framework has not materialised. In banking, the use of generative AI has the potential to reduce the marketing spend.
Coforge has developed an intelligent data processing solution called Quasar. It is layered on top of ChatGPT for their insurance client.
Generative AI is useful for customer services and IT service desks.
Indian IT forms are building their own prototypes. They build federated AI which allows the algorithm to learn from several different data sets without exposing them to each other.
There can leveraging of the capability to build solutions. Alternatively, there could be a consortium using the concept called federated AI. Open source generative AI so solutions tend to require more computing resources. It is cost effective for enterprises to opt for algorithms that work on limited datasets.
Though generative AI is welcome, the delivery should be free of any infringement, and full ownership of IP.
Microsoft owned GitHub and Open-AI has launched Co-pilot in October 2021. It is a solution that generates code. Google has launched Bard. Microsoft has infused ChatGPT with Bing, Edge browser and Dynamics 365. There are concerns about data ownership of source, data integrity and general accuracy of responses.
Infosys is exploring use cases across open source as well as proprietary generative AI and LLMs.
AI has become a part of the training package.
Generative AI security concerns make the employers hostile towards ChatGPT, asking the employees to be cautions.
IBM has launched IBM Watson X enabling enterprise to scale and accelerate the impact of generative AI with reliable data. It will make the enterprise uses safe and secure.
Some governance mechanisms can be built into the algorithms. There could be filters to pass the data. These will check hate, abuse, profanity, language duplication, copyrights. There could be a series of benchmarking. The right use cases should be selected.