Last week we issued our article on GenAI !
https://lnkd.in/eFQN5vzC
Before discussing use cases, GenAI is still vague for many people so here are 7 elements for an overview :
➡ Core Purpose : Creating new content, not explicitly programmed, from patterns learnt.
➡ Models : It relies on deep learning models, and commonly utilizes unsupervised/self-supervised learning. In other words, data is given to the model without human tagging, and model creates the 'logic' rationale itself.
➡ Dev Timeline : As it's linked to Deep learning, GenAI originates from the 1950s, but many breakthroughs occured in the past 5-10 years. Or at least the breakthroughs that led to the current hype.
➡ Industry Adoption : Full-scale enterprise deployment is largely experimental, though many uses cases are being tested. We see it everywhere at clients, but in my point of view it will need few years to see the industrial breakthrough at scale
➡ Explainability & Transparency : Often functioning as 'black box', without understanding how the models got to the output. Outputs may also be inaccurate, inconsistent or non sensical. Those all create challenges for enterprise industrialization
➡Scalability : GenAI scalability is limited by high infrastructure and energy requirements. Energy demands are so high that GAFAM are pushing towards more nuclear plants in the US to meet that demand. The companies with operations in cheap energy countries will have an advantage (like in the Industrial sector)
➡Regulation and Governance : Regulatory frameworks are emerging to tackle transparency, explainability, content ownership and bias issues. There is a high risk that regulation hinders innovation by creating additional complexity, lag-to-market and costs. Companies with the ability to operate and sell in markets with softer regulations will take the advantage.
What are your thoughts on it ? ⤵
BG&A (Blanc Gay & Associates), #GenAI, #ITFM, #IT, #AI
