Our is going no where fast. Promised returns and automation delayed. We always seem to be getting different results from the same prompt. Still. AML part is going to be pushed live, fake it til you make it right?
8 replies (most recent on top)
I am steering clear of their AI. It’s just another way for them to surveil, steal our work, plus collect our data and prompts to train their own systems. Which ultimately serves their layoff agenda.
I asked AI what to write for my mid-year, copied it verbatim, and pasted it into Workday. 30 seconds. My manager and the MD will use AI to read it then AI to rank me. They will then use AI to determine the minimal amount they will raise/bonus me next year (2026 is rich for our biz). Otherwise I have no use for the AI which could cover 90% of what I do. The other 10% ain't worth 1/5 what they pay me. RIF me, please....
creating html dashboards, manipulating excel (i.e, joins, look ups). It's gotten pretty good at data analysis. create a 1 pager of my analysis I can send to a senior, etc.
I am using it to create PowerPoint decks using multiple source documents and policy documents. What took me over a week in the past now takes about an hour to create, then another hour to edit all of the hallucinations out. Its also scary good at comparing spreadsheets and spitting out the differences between the data in them for further analysis. Instead of going line-by-line through hundreds or even thousands of rows, I get all the data I have to look at more carefully in about 5 minutes. It makes conditional sorting and vLookup functions seem like stone-age technology.
AI, specifically Citi's Copilot gives me great concise answers to questions that Legal and Compliance gave me wish-washy answers on. I knew I was right...
My AI efforts consist of using AI to pass the Citi training opt-out pre-test, every time. Lots due this week, CoPilot will be busy (Grok on the side PC even better)
whatever was left in NY , TX or FL after rounds and rounds of outsourcing are now being pushed to India anyway. they hire two- tree VPs in India to replace 1 VP in the US. yes, you have to pretty much to spoon-feed and extensively train those team members - luxury never available for Americans - but sooner or later those jobs will be in India. yes, quality will be much lower due to multiple reasons including fast turnover, but that's the risk company seems to be willing to take. An attempt to marry AI with cr-ppy data and business apps produced and maintained in India might very soon result in yet another consent order.
Would it be fair to assume that those activities which lend themselves most readily to Agentic AI adoption were sent offshore a long time ago?
That leaves stateside processes that were kept in NYC/NJ for a reason. They are more complex, or they require sophisticated business knowledge, or they call for nuanced human judgement, or some combination thereof.
Now the banks are trying to turn their commercial lenders and KYC auditors into prompt engineers and LLM trainers, with minimal training and unproven open-source tools. So how is that working out? Uh, not so good I take it.