Will AI Take Your Job? Deconstructing Job Descriptions To Understand What Roles Are Most Vulnerable
Like it or not, we’re seen through the lens of our job description. It defines the tasks we complete to create value for the business. They will also set the bar for which roles AI replaces.
In this article, I will break down job descriptions and discuss what AI can do today, what it will be capable of within 2 years, and what’s safe from automation. You’ll see that some roles are so poorly defined that executives will believe they can be automated.
However, what exceptional employees really do isn’t captured in most job descriptions. Job descriptions are written for HR screening and job interviews. The capabilities required to succeed in each role are often very different.
On the other hand, many jobs considered critical today don’t deliver much value or require complex capabilities. The better businesses understand how people create value, the more these roles will disappear or be automated. The AI Product Manager is a good example, and that’s where I’ll start.
AI Product Manager Version 1
Lead strategic planning, goal setting, and roadmap development for the conversational AI product and the LLM foundational libraries.
This requirement is more significant than a single line can capture. Most of an AI Product Manager’s time should be spent on this piece. However, if you read between the lines, this company doesn’t put much emphasis on it. Summarizing it this way instead of defining it more completely is a clear giveaway.
GenAI can’t automate this part of the job, but that only matters if it’s the core function vs. an afterthought.
Maintain a comprehensive understanding of LLM use cases, facilitating connections among individuals addressing similar challenges and guiding them toward solutions that leverage reusable components or common methodologies.
An LLM can keep a database of use cases and perform a similarity search to match new requests with existing solutions. It can provide information about existing solutions and help users select the best tools, components, and methods.
This is a moderately complex recommendation use case. It’s a variation of what companies like DynamicYield have been doing for years. Amazon recently introduced personalization and product descriptions using GenAI. Adapting from retail to supporting this use case will happen in 2 years or less.
Define and champion a strategic vision, acting as the primary evangelist for this vision and ensuring the product meets risk and control requirements.
GenAI can create a template for a strategic vision or a very generic one. If that’s all the business is interested in, that part is at risk of automation. Consulting companies are already using AI to build strategy documents and customize templates.
Developing an AI platform strategy and vision is beyond where GenAI is or will be any time soon. Those should be heavily customized and connected with the business’s strategic goals. If the business is interested in more than just the document, this part is safe from AI.
For now, the champion and evangelist roles are safe as well. As conversational agents improve, they will become decent evangelists. They never lose enthusiasm and have as much time as customers or stakeholders need.
Engage with senior stakeholders across various businesses and functional groups to collect business requirements and create detailed product requirements documents.
GenAI easily takes over requirements gathering. I saw it at Dreamforce. Customers sat down and explained their problems. Agentforce built solutions based on those conversations. AI will rapidly take over the requirements gathering and documentation work.
Oversee and prioritize high-impact initiatives requested by stakeholders, ensuring alignment with the strategic vision.
This translates into managing changing priorities and handling fire drills. Moving tasks around isn’t very difficult. This will be automated in less than 2 years.
Collaborate closely with technical teams to direct and oversee the planning and execution of the product roadmap and use cases.
Sprint planning, backlog grooming, and setting deadlines will also be automated in less than 2 years by the same tools as the last line. This is really low-end work that a GenAI agent can perform with access to apps and data sources.
Provide comprehensive product and use case status updates to senior management and stakeholders.
Again, an agent can handle status updates with access to data and apps. It can handle the emails, collation, reporting, and distribution of status updates. This will be automated in less than 2 years.
AI Product Manager Version 2
The last job description was from a traditional industry, and this one is from a tech company. The differences are immediately apparent. This version of the role and people in it will be much more resilient to automation and AI. The challenge is that people with the job title and version 1 responsibilities aren’t qualified for the AI Product Manager roles that will be available in 2 years.
The capabilities they spent years developing won’t be in high demand. It will be a gut punch to go from a high-paying role to being unable to find work. I recommend that if AI can handle more than 30% of your job, it’s time to upskill.
Display strong leadership, organizational, and execution skills.
GenAI will help AI Product Managers be a lot more organized. NotebookLM and scheduling assistants are good examples of how AI will manage the most challenging parts. That will be significantly easier in 2 years. Strong leadership and execution skills are not in danger of automation any time soon.
Is the primary driver for identifying significant opportunities and driving product vision, strategies, and roadmaps in the context of broader organizational strategies and goals.
Opportunity discovery will be AI-enabled, making AI Product Managers much more effective. Replacement is well out of reach.
Framing product strategy and vision in terms of broader organizational strategies and goals will also be AI-enabled. The bigger the company or the more parts of the company that the product touches, the more complex alignment gets. GenAI makes a great sounding board that doesn’t get tired and tracks constraints so they aren’t overlooked. AI assistants will move from prototypes to products in less than 2 years, so help is coming.
Incorporate data, research, and market analysis to inform product strategies and roadmaps.
This will be AI-augmented as well. Product Managers will be a lot better at research, and it won’t take up as much time as it does today. Research often gets shortchanged because there’s insufficient time to do a thorough job. In 2 years, GenAI will handle most of the grunt work, and AI Product Managers will read the tea leaves for insights.
Leads and motivates a team of engineers and other cross-functional representatives and maintains team health.
Nope. GenAI won’t take this over any time soon.
Understand {company’s} strategic and competitive position and deliver products aligned with our mission and recognized best in the industry.
Competitive analysis will be supercharged by GenAI the same way other market research will be. The challenge with getting a sneak peek into the competition’s playbook is finding its data leaks. GenAI can scan through more data sources and surface potential leaks. Staying ahead of the competition will get easier in 2 years.
Maximize efficiency in a constantly evolving environment where the process is fluid and creative solutions are the norm.
Optimization problems are a machine learning strength. In 2 years, we will rely on models to identify potential improvements and ways to streamline operations or processes. AI Product Managers won’t be removed from the loop, but this will be mostly automated.
Senior GenAI Strategist
You will partner with customers and 5 LoB delivery teams to craft industry-specific programs to serve customers' needs and extend that use to other customers within the industry and related industries. You will leverage and refine a programmatic approach to help customers drive their business metrics across a set of industry-specific needs. As a trusted customer advocate, the Generative AI Strategist will assess and deliver best practices around program delivery, project approaches, and business measurement. The ability to connect business value to a scalable, repeatable, extensible approach is critical to the Strategist role.
I spent 10 minutes trying to find something that could be automated and there isn’t any part of this AI will take on. It will facilitate communications at scale and support summarization, but that’s it.
Bring alignment between customer’s technologists and business leads; help them to explore the art of the possible with generative AI and machine learning, and to develop a roadmap to deliver business value most effectively.
While evangelist and advocate roles will be heavily supported by GenAI, the way this is written points to a much more mature vision for those roles that can’t be automated. It’s more than getting people excited and teaching them what emerging tools can do. Alignment between the business and its customers isn’t something AI can take over.
Discuss complex industry-specific business concepts with executives, line managers, and technologists.
This is an excellent GenAI use case for some conversations. However, there’s no replacing a person for more nuanced discussions of AI’s implications and opportunity selection. People on the front line want to talk to other people who get their problems and are committed to translating business needs into technical solutions.
Engage with various internal teams, including applied scientists, solutions architects, business development, marketing, industry specialists, partners, and regional organizations.
Most of the engagement will be automated, but the translation will not. Translating between teams with different domain expertise and aligning them with shared goals isn’t an AI use case I see as feasible within 2 years. I think this happens eventually, though.
Drive large, complex sales opportunities from ideation through vision building and scoping all the way to closure and into delivery.
Nope. GenAI can make a capable salesperson, but enterprise customers or consumers making major purchases don’t want to be pitched by a machine.
Conduct workshop sessions to identify opportunities with our customers to scope how they could deliver business value through the use of generative AI or machine learning.
Conducting the workshops will be 100% people-driven for the next 2 years. Everything else will be AI-augmented. GenAI is exceptional at reading multiple workshop transcripts to find overlaps and themes. It won’t be up to a person remembering, reviewing notes, and synthesizing what they heard for much longer.
Java Developer Key Responsibilities
Design, develop, and maintain complex software systems and applications.
This is the grand debate that’s raging. AI can write code, but the quality is inconsistent. It can recommend lines and entire code blocks. Again, recommendation validity is inconsistent. GenAI tools are improving very quickly. OpenAI’s Canvas and Cursor show how far we’ve come in 2 years. In the next 2 years, we will see even bigger improvements.
However, we all know there’s a plateau coming for the current crop of foundational models. I don’t think we have a good sense of when or where the coding capability ceiling is. I spent over a decade in software engineering, and after using Cursor, I’m worried about the future of that profession. I don’t think a 100% replacement or automation solution is coming. I see AI taking on over half of the code being written in 2 years.
When libraries and frameworks entered the software toolkit, the impact was an acceleration of demand for software engineers because the unit economics worked for more projects. Software engineering will see a similar boost with AI coding assistants, making more projects and products feasible. GenAI also makes more projects feasible for non-developers to complete.
Agent builders are designed for non-developers. Agentic systems can be built without software engineers, data scientists, or data analysts. Developing the tools and data pipelines agents need to complete their work is also increasingly automated.
We will have a clearer picture in a year. I think this requirement will be largely automated in 2 years, but never entirely. Software engineers will continue to play a role, but it will be more like what IT does today for systems and infrastructure.
Collaborate with cross-functional teams to gather requirements and define technical solutions.
As we discussed earlier, GenAI easily takes over requirements gathering. Interfacing with external teams and understanding their needs is a different story. Most software engineers do the former but not the latter.
Implement and maintain best practices for software development and engineering processes.
GenAI coding tools will take over enforcing best practices and optimizing code. GenAI is good at refactoring, optimization, and security recommendations today. In 2 years, it will manage a lot of this work.
Develop and maintain software documentation, including design specifications, user guides, and manuals.
AI will take this over, and no one will shed a tear over it.
Ensure the reliability, scalability, and performance of software systems.
This will be AI-augmented but not automated any time soon. Companies like AWS use models to understand how changes will impact the rest of their increasingly complex platforms. AI will show software engineers downstream impacts and upstream threats that would take a person days to find.
Troubleshoot and debug complex software issues.
GenAI tools make excellent rubber ducks for debugging. They will probably take over most of this work within 2 years. Again, I don’t think anyone will miss it, but this is what lowers the skill level required to build apps. If GenAI can recommend code and clean up defects, many more use cases become feasible for self-service and low-code platforms.
Mentor and coach junior engineers.
Nope. AI tools will make junior engineers more capable and accelerate their learning curves. Nothing can replace 20 years of experience and the ability to quickly transfer knowledge so junior engineers perform at higher levels.
Data Scientists: A Growing Threat Of Commoditization, Not Automation
Buy vs. build is leaning against data scientists. Most teams primarily support internal use cases. Traditional SaaS vendors are pivoting hard to a hybrid AI business model. They use the datasets built from years of customers working on their platforms to train models that make users more efficient. They have better data and economies of scale on their sides.
Foundational models are so expensive that business leaders have overwhelmingly chosen to buy vs. build them internally. Costs for even the most significant models are falling rapidly. Between platforms that provide access to foundational models and private instances on AWS, Azure, GCP, etc., the benefits of GenAI are accessible without the headaches and costs.
If foundational models and AI supporting internal operations are increasingly commoditized, what’s left for data scientists? I have talked with several data scientists and executive leaders inside of tech, finance, and retail companies who see less work coming down the pipe for their teams. One said, “It’s hard to get them to see the writing on the wall.”
AI engineering and applied machine learning are increasingly in demand, while model training and theoretical research aren’t. A student in my current AI Strategist Certification cohort said, “I could make a case for buying everything.” I agree that internal operational use cases will be 80%+ buy decisions. Products are a different story.
Most companies haven’t figured that out yet, and data teams must drive that realization. Big tech sees AI as a growth driver, but traditional businesses don’t. There will be a lag, but AI products and features will be just as critical for traditional businesses.
Data and AI are new inputs for growth. Customer expectations will change as more AI products reach them. Put the two forces together, and the trend is inevitable. Data and AI teams must pivot to customer-facing product teams to remain relevant.