Don’t Use LLMs to Make the Boat Go Faster… Head for Space!
Like many, I've been experimenting with Large Language Models (LLMs) to gauge their impact on the businesses I work with. My key realization? Using LLMs just to speed up our existing 'boat' risks being left behind by 'spaceships.'
The Paradigm Shift with LLMs
The emergence of LLMs like GPT-4 isn't just an evolutionary step; it's a revolutionary leap, akin to transitioning from sea navigation to space exploration. These groundbreaking technologies offer more than mere enhancements; they unlock a realm of unprecedented possibilities. We're not talking about marginal gains here; it's about a bold redefinition of what's possible.
The Real Challenge: Shifting Mindsets
However, I've learned that the true challenge with LLMs isn't in their sophisticated algorithms or vast data sets, but our capacity to break free from entrenched processes and conventional thinking. To truly harness LLMs' potential, we must fundamentally shift our focus towards objectives, rather than merely improving upon previous solutions.
Recommended by LinkedIn
My Approach: A Framework for Thinking Different
Here's the framework I’m using to guide the teams I work with:
In conclusion, our journey with LLMs is not about incremental improvements to the boat, but about launching a space rocket that completely redefines our approach to problem-solving. By embracing this transformative mindset, we open ourselves to a future brimming with untapped potential and innovation.
And yes, it's incredibly fun!
Reimagining experience and service using data and GenAI
1yGreat read. Completely agree with the idea that the conversation should be outcome-centric and that focusing too much on how we do things now can slow our thinking down.
Connecting with brands and partners who want to humanize the customer experience with VML MAP
1yInsightful post, Matthew! Fully agree, LLMs are about redefining possibilities, not just improving existing processes. Exciting times ahead!
Founder at Orchestra
1yThis is such a good point. LLMs are fundamentally suited to a very different set of problems to the ones we are used to solving. Great stuff Matthew Tod
CEO at Snowplow — London, Boston
1y100% agree. I think there is an assumption that LLMs will slot in as kind of virtual task doers in existing processes, but I am seeing them having far more leverage than that as strategic thoughtpartners to come up with new paradigms…
Enterprise Software Sales Leader | 20+ Years in Martech, Adtech, Data, Analytics, AI/ML & Cloud | Driving Multi-Million Dollar Deals & ARR Growth Through C-Level Partnerships.
1yThanks, Matthew, reading your posts is always insightful! Appreciate you!