I would like to seek leaders opinion on - what are envisaged updates/ advances in Gen AI in next 12-18 months ? How best could a leader foresee the deployment of Gen AI tools in asset and wealth management in next 12- 18 months?
Chief Strategy Officer, Self-employed
Would it be towards better customer experience and enhancing the ongoing and future business process flows? Chief Data Officer, 2 - 10 employees
One Gen AI Advancement I expect isProgram-Aided Language Models (PALs),
Can improve the reasoning capabilities,
Of large language models (LLMs).
PALs work by generating programs as intermediate steps for reasoning.
LLMs focus on the decomposition of the problem into executable steps,
Using Chain of Thought, and Few Shot learning techniques.
While the actual process of solving the problem,
Is handed off to a programmatic runtime,
Such as a Python interpreter.
PALs have been shown to be effective in a variety of reasoning tasks,
Including arithmetic and symbolic reasoning.
And are more accurate than traditional LLM methods.
While PALs are still under development,
They have vast potential to enhance LLMs reasoning.
And can be used to automate a variety of tasks,
Such as solving math problems, writing code,
And generating natural language descriptions of algorithms.
PALs could also be used to develop new educational tools,
Such as interactive learning environments,
That help students learn how to reason and solve problems.
Attached is research from Carnegie Mellon University,
Detailed the method to improve LLM reasoning,
With generated programs.
https://arxiv.org/abs/2211.10435
Chief Strategy Officer, Self-employed
Thanks Dan for sharing your insights and the research paper. PAL looks promising !
Chief Data Officer, 2 - 10 employees
you're welcome
Chief Strategy Officer, Self-employed
Would you consider 'finetuned niche models' to make an impact ? Chief Data Officer, 2 - 10 employees
yes models fine tuned on specific knowledge domains along with retrieval augmentation will help address hallucinations.
Chief Strategy Officer, Self-employed
Indeed, any thoughts on how should teams prepare for these deployments ?
Director of Data and AI in Banking, 10,001+ employees
I think wealth management could see benefit from all the key capabilities of LLMs...content generation, summarization, and semantic search (probably not code generation). I like to think of it in terms of those capabilities. I could use it to prep the advisor for client engagement by gathering infomration tailored to the customer, generate a powerpoint based on a template (with M365 copilot). I could summarize long documents into customer facing Q&A materials. I could search for companies that have non-standard offerings or things that are difficult to find in published materials. This type of usage seems to change enough that I would image the retrival augmented generation with vector databases would be the way to go, I don't see a universe where you could build a model fast enough (or cheap enough) to tune all the info necessary into the model.
Other thoughts for asset and wealth management...Building a client profile from social media and other sources, building a client profile, using that profile to recomend portfolio optimization, summarize client meetings for regulators or updates to CRM.
Chief Strategy Officer, Self-employed
Thanks Jeremy, I agree with your points. I spoke at an AI event recently and it was heartening to learn that the leaders in the field are warming up to the idea of running some pilots for internal stakeholders in marketing and sales domains.
Content you might like
Data mesh41%
Data fabric41%
No12%
Don’t know6%
17 PARTICIPANTS
CTO in Software, 201 - 500 employees
Without a doubt - Technical Debt! It's a ball and chain that creates an ever increasing drag on any organization, stifles innovation, and prevents transformation.Yes, increased headcount18%
Yes, reduced headcount47%
No, expect it to stay the same24%
Not yet determined12%
I don’t know0%
17 PARTICIPANTS