Financial Services Cloud Council and Forum

for those who do not use LinkedIn - 

NEED for AI governance - yesterday at the https://lnkd.in/er8VYk3N New York Briefing - I raised a question on the risks and future of alternative hashtag#data and the current hegemony and walled gardens being created by big tech. Argyro (Iro) Tasitsiomi, PhD replied with a few additional comments, and she has an interesting article (https://lnkd.in/ej_JnCVb) about the alacrity of global data growth and how despite that massive data generation there are implications for the potential "exhaustion" of data as early as 2026. We have already seen the first wave of this tech exhaustion when the big tech LLMs hit trillions of parameters off the available "quality" of content globally and started "leveraging" (through transcription of voice and video) from content publishers. These are not just reputation and lawsuit risks but fundamental governance issues.

One of the big themes across NY hashtag#Techweek was how content hashtag#creators and brands could protect themselves and their content and the hashtag#sustainability and hashtag#ethics of current data-intensive approaches to LLM training. It behooves the industry (especially FinTech) to take a few steps back and remember that LLMs do not have logic, much less souls and ethics. Models are created for a purpose - for the most part to simplifying complex real-world situations and workflows by surfacing underlying patterns - which means that they are purpose built and the optimal relationship between corpus, model size and the best performance either from scale, efficiency or precision and recall is defined by purpose.

The bigger ethical and governance questions - especially with the potential hegemony through scale and training and a tyranny of the data majority through eyeball/mind share is an AI governance issue that needs awareness and much bigger issue than hashtag#modelbias or hashtag#explainability or any emphasis on data-to-model ratios or model efficiency. We not only need to rethink our approach to model development but understand AI governance and Data governance in the context influencing factors that require fundamental governance as well.

the LinkedIn  because some links did not come through : https://www.linkedin.com/feed/update/urn:li:activity:7204828046372040704?updateEntityUrn=urn%3Ali%3Afs_updateV2%3A%28urn%3Ali%3Aactivity%3A7204828046372040704%2CFEED_DETAIL%2CEMPTY%2CDEFAULT%2Cfalse%29&lipi=urn%3Ali%3Apage%3Ad_flagship3_profile_view_base%3BlXCzqLIUR9GL9QW081scpw%3D%3D