AI – the buzz word in the world. On the surface of it, everyone is so excited about not having to write for SEO, get their chosen model to do their tasks or have Agentic AI. I am not excited about this. Those types of processes that are now just glorified workflows are more glitzy on the surface than real on the backend (yet). Also, it is what generates headlines because everyone can relate the slight fear sprinkled across discussions about AI.
I want to talk about the AI features that will change our lives for the better. One of the first podcasts I listened to on AI years ago was an early interview with the father of AI I believe. Talking about ChatGPT becoming a mainstream discussion point back in 2022/3, and what stuck with me back then was him talking about the implications of AI in healthcare. How it can change early diagnosis of cancer, spotting patterns the human brain is just not as capable of doing yet. It fascinated me. A world where technology accelerates our wellbeing. However, as every story, the world did not pay attention to these maybe niche conversations and all everyone has been talking about the last 3 years in the mainstream media – customer service agents, self driving cars, sales funnel management and website/workflow automations.
My personal view – if we are going to burn through tokens and electricity, as well as water supplies may it be for projects that change the world and humanity for the better. I do not know why would I get excited about John in customer service not having a job in a year because AI has taken it. I do not see how this is a better or more efficient world really. The Citrini research published last weak paints a bleak picture of the world just 2 years away if this is our main focus and narrative going forward.
Veraia Life is my product, so I understand you may think I am biased. But hear me out – with this development I am not looking to replace humans anywhere in the chain of command. I am looking to create an easier, more streamlined journey for the human behind the screen. Allow for confidence in moments of high emotion and stress. My vision has always been that AI should be here to support our lives not hinder them. However, in order for this to be true, we need to be able to understand AI’s chain of thought, or in other words – why is this output this output. Why is it when the same 2 propmts are asked, they get different answers, why is it that over time outputs lose accuracy. Without these questions front and center we are just riding a ave of hype with no guardrails.
How does this tie into AI observability?
Observability is expensive, in a token world. This is why many companies developing technology, sit on the fence of calculating ROI of observability versus output. They count tokens and the economics has to work and this is the main pillar of their output.
The reality for us at Veraia Life is that observability is non negotiable and is a core part of our development vision. We want to know why an output occurred, we do not want to guess or assume. Our development team is meticulous in having 100% understanding of the CoT (chain of thought), as in why a certain output got produced every step of the way. We rather sacrifice speed of analysis and output, than accuracy. This is what good UX is for in an app world – we keep our users engaged until the correct output is displayed.
And yes, this means human supervision on factual correctness for the output. Although most AI businesses love to boast numbers of autonomous decisions, the reality according to Dynatrace is that 69% still require human supervision – and you know what? That is OK. I am a believer that AI is a supportive co-pilot which enhances areas we humans are weaker – speed and pattern recognition. But in my humble opinion, we are still far from autonomous AI systems like in a sci-fi movie. I believe in a grounded approach that is rooted in security, consistency and constant iteration and adaptation to serve the best possible output for our customers.
I do not want Veraia Life to ride on buzz words. I want it to ride on trust in its operating model and output decision making, even if that means that autonomous decisions come further down the line. A product based on these principles takes time to build but we are committed to our ethics and values, so we can deliver something truly useful to our end user.
