Thoughtly’s Post

“It’s missing XYZ.” Yep, I totally agree. I had another probably 10 rows that didn’t make it in (including KPIs, ethics, in-house / outsourcing / partnerships). Maybe a v2 when I’m not exhausted from flights. Thank you for the ideas 💡

How does one "test new free consumer tools safely" - it is abundantly clear that very few organizations understand how (if at all) LLM's protect data, prevent infringement of IP, prevent bias and hallucinations, the introduction of malicious code and the propagation of fake information deliberately introduced. Similarly, many corporations still have not developed sufficient maturity in basic IT Security practices (can anyone say encrypt data at rest Evolve Bank & Trust), so expecting robust policies on AI use and expecting curious employees to know how to "test safely" seems unwise. Accessible and usable AI has tremendous potential, but we need a massive leap forward in collective understanding, proper governance and adequate controls before we take the training wheels off.

Why limit this to just GenAI? AI in various forms has been adopted by many companies for around a decade. GenAI is just the last tool in the box, but companies shouldn't be fixated on using it just because its there. In many cases, NLP might be a better/cheaper solution than an LLM. The important part is to identify the problem and test which solutions fix it, bearing in mind data access, tooling, governance and the ability to efficiently maintain it in Production.

These dimensions broadly align with what we’re seeing across Microsoft‘s largest FSI customers. I think two key dimensions should be added to this overview: 1) Separate “Tools” into apps, platforms and models. MSFT sells all three and there’s a different maturity profile for each. 2) Partnership strategy is a key component of the maturity journey. Every firm needs a framework for how to think about build/buy/partner in this rapidly evolving landscape. Those that don’t will experience significant pain when the inevitable shakeout comes.

This is a great chart but the Gen AI journey in large companies has an additional twist. Even within top leadership, there is no mindset alignment. Some leaders have the behind mindset, some the on track mindset, and some are true believers and are working to make it happen within the realm of their business unit. This is particularly common in companies with dispersed analytics and modeling capabilities. Truly ahead only happens when a company's entire top leadership believes in the transformational power of Gen AI.

I have a differing opinion (shock!) in that most of this chart isn't reality. This is the epitome of the AI desperately searching for a problem it can solve. It's focused on generating AI FOMO, not evaluating the value of AI. Which checks out, because the narrative here is the AI is *inherently* valuable, but the data outside the hype just doesn't support this. Implementing AI is time consuming, harder than you think, and more expensive than you are expecting. Instead of assuming you are "behind" instead ask yourself: * Does my business actually need AI? What problems am I looking to solve that can be solved by AI? Can I actually get value out of it? It's a high risk, high cost, medium reward endeavor - do I really need it? Most of the rest of this checklist doesn't matter if that answer to that question isn't yes. And spoiler friends - for 99% of companies out there, the answer to this is No. Making bombastic statements like "AI is a shift in the way we work" is the hype talking, advocating for "reallocated headcount" (which is a fancy way of saying firing some folks to hire others who probably lied about about having AI experience on their resume) manages to undervalue your existing staff and overvalue the benefit of AI.

Dig the emphasis on needing to incorporate AI deeper into the org. Plug: we just released some free thinking/facilitation tools to help with that: https://nobl.io/changemaker/ai-strategic-vision-toolkit/

Motley Fool is hedging heavily against GenAI so take all their employees opinions with a grain of salt, and 2 shots of tequila. Takes like 5 min of research to figure that out.

I would add to "Ahead": 1) have a governing body that allocates funds to AI projects based on impact to bottom line. 2) have a technical oversight group that coordinates tech stack and architecture for AI projects. Too many big companies have 500+ disconnected AI projects that are duplicating efforts and not at all integrated. Or have no real business case to justify the effort / costs.