Jobs AI can't replace
Rather than giving a list of roles that AI can't do, I believe it's more valuable to give a list of principles and ideas that you can implement at your workplace to work alongside AI.
Let's understand what work is.
Work in the modern day can be classified into 2 types, Valuemetric and Volumetric.
What does this mean...
Volumetric
Work that benefits from repetition with minimal changes to the required outputs, most of the work that we do in our lives is like this, think folding clothes at a personal level to marketing, sales and coding at an industry level.
This work benefits from our ability to be smart, through retaining knowledge and executing ideas based on the limited context that is required for the job. Like accepting or deferring a loan application, there is a limited set of data (credit score, salary, loan amount etc.), and based on a calculation a binary decision is made.
Coding is similar, there is a limited set of tools that can be used, the problem that needs to be solved is objective, then these tools and services are connected through code to achieve said objective.
You get better at this style of work from doing more volume, a person who ships more features in software is paid more than someone who ships less, a loan officer gets promoted when he processes more loans than his peers.
Valuemetric
This is a style of work that doesn't usually benefit from volume of work done, rather it depends on how many dots you can connect.
For valuemetric work , you need a narrative, a story, a why it is that you do the work. What type of value are you providing to your business/clients/team? This type of work is akin to what a lot of founders do day-to-day. Understand what is valuable in the marketplace, is it saving time? saving money? lowering code complexity?
This style of work has typically been reserved for the founders and CXOs, and other high-level executives. but now, this type of work will be more valued then ever before.
You get better at this style of work from understanding the power of value, it's not about building a product for the sake of the product, it's about understanding the world around you and understanding what values you and your customers align with and build accordingly.
If it isn't evident so far, volumetric work is something that LLMs specialize in, since their job is to predict the next word not what is most valuable. LLMs, especially with Agentic harnesses have the ability to do this style of tasks at a volume humans cannot compete with. Valuemetric work is something a machine cannot understand since it doesn't have a system to leverage emotion, values and beliefs like people do.
That's why no matter how good these models get, they are awful at storytelling, because a piece of art takes decades of human experience to express on an instrument, and can't be understood by matching patterns.
What are the steps you can take to work with AI
Step 1: Learn how to leverage the volumetric workers (AI Agents)
Understanding the basics of what an AI Agent is and its parts (Memory, Model, Tools). And spending time and money on infrastructure to run these agents: this can be anything from an old PC with an AI subscription to a $50,000 server. Then setting up tools so that your agents can run actual workflows.
Step 2: Split the work you do into 2 buckets, Valuemetric and Volumetric
Based on your job, you might be really surprised either way on the type of work that is expected from you. Align your volumetric tasks with agents and hyper-specialize on valuemetric strengths in your work environment, because THAT is your human leverage.
Step 3: Execute, Repair, Repeat
The agentic systems we have today are really advanced, but they are still prone to mistakes, upgrade to different harnesses, newer models, better connectors to your services. And overtime as value at your organization changes, learn to keep up with them. Like any job, you are constantly learning.
These are my thoughts on the next generation of the workplace, where you have thinkers and creative individuals who build the value proposition of an organization and instruct agentic systems to execute volumetric tasks.
Comments
Post a Comment