I'm worried that AI may replace my job as a developer; are PM jobs safer?

Everyone used to say that AI will aid employment rather than replace it. What I can do in a week, GPT-4 can achieve in a matter of minutes.

Since a PM’s position is a generalist one that requires communication and planning. I believe AI will support PMs in their work, not replace it.

On the other hand, I believe AI won’t replace top level technical architects or tech leads, but rather software developers who are essentially performing ‘code-monkey’ type duties. That implies that you would only require two engineers as opposed to four. You would still require a product manager, though.

Have you used GPT-4? What do you think about it?


Putting on my PM hat, I’ll argue for twice as much work to be done if there is tooling that enables two engineers to perform the duties of four. Most non-trivial initiatives have the biggest bottleneck during the engineering phase. ML tools help in improving that, right? Oh, yeah. My developers can now perform more.


While in certain instances this is true, it is not always the case. For instance, just because you triple your efficiency doesn’t necessarily guarantee that your productivity would triple across all teams and organizations. Some people can notice a 3x improvement in output, while the average improvement might only be 1.3x or 2x (this is still unknown and will fluctuate over time).

In addition, if everything else remains the same, you would actually require more PMs in relation to engineers.


Yes, a logjam could emerge that there’s not enough product designers. Or PMs. Or customer success. Or sales. I guess my point is that this feels to me like it only increases demand in the software sector.


At a high level, every job role comprises three components:

  1. Technical talents include the use of tools and hard skills like coding, html, CSS, SEO, design, and photoshop, among others.

  2. Experience in a domain, industry-specific knowledge, and market intelligence

  3. Human qualities include coordination, empathy, communication, and leadership.

One issue that people frequently overlook is the necessity for someone to oversee the input and output of AI, even if it partially replaces the technical faculty.

For instance, in the case of front end development, someone still needs to collect requirements, feed inputs to AI, iterate to get it perfect, then take the output to a stakeholder, obtain feedback, connect with other teams to launch into production, etc.

Now, how a front end development function falls on the three criteria listed above will determine whether it is in danger.

An AI product will be a threat if you’re just taking a ticket out of JIRA, putting it in a folder with minimum human involvement, and using quite basic site designs. The same will happen to transactional services like transcription, translation, image categorization, etc.

Your employer will still require you to manage relationships if you are providing complex front end executions in addition to human coordination. Your use of AI to streamline menial tasks is likely to be expected, and productivity benchmarks may change.

There is an indirect threat in this situation since businesses will realise they might not need as many front end developers to complete the same amount of work. Moreover, they might not need “front end specialists” either if the task’s output is rather simple. There will very likely be prompt engineers.

Roles where there is a requirement for someone with deep subject expertise, experience, and knowledge of undocumented previous history of business/customers, in addition to technical and human factors, won’t be threatened in the short to medium term. They will undoubtedly be need to employ AI as well.

Employers will soon start making efforts to democratise that knowledge over time through AI in order to stop depending on human databeds or important personnel whose departure could jeopardise operations.

Knowing how many of these levers your work pulls is crucial. Workers that are adaptable will look for opportunities to make the most of their domain knowledge and human attributes. They will also decide to shift their attention back to difficult technical tasks where there will probably be little training data available.


Thanks @EvaRichardson! Your answer shows critical thinking on this subject and presents valid points instead of blindly rejecting the idea that AI may reduce the need for certain skillsets of software engineers.


Is there not a way to give the AI system a more closed loop in relation to the input/output model?

In my mind, the success metrics would be developed and fed back into the system by input from the analytics tools via some kind of automated script that pulls the most recent numbers and inserts them into a query to have AI generate new criteria based on the high and low performance across key metrics. The acceptance criteria themselves might even be written with natural language processing. Next, repeat the process.


In my opinion, the most likely scenario is that software development will change but not be replaced. The same is true of numerous other vocations.

Software development is similar to PM in that what you do at a SaaS B2C business in California is very different from what you do in a pharmaceutical company in Switzerland. However, PM is more difficult to replace.

Imagine that these businesses decide to replace SW engineers with AI, and no one is able to check every line of code. Can you picture the commotion if a user were to experience something? (Patient)

Trains, finances, automobiles, law, space, and defense.

The position you hold at your company may be in jeopardy, but software engineering is not.


Indeed, programmers once wrote in assembly. Coding was only made more powerful and user-friendly by abstraction, which increased the complexity and creativity of development. The same scenario applies here.


I can provide some insights, but please note that predictions about the future are uncertain. The impact of AI on different job roles can vary, and it’s challenging to make definitive statements about job security. However, I can offer you some considerations regarding the role of Product Managers.

While AI has the potential to automate certain aspects of software development, including code generation and testing, it is less likely to completely replace the role of a developer in the near future. Developers bring creativity, problem-solving skills, and a deep understanding of software architecture, which are not easily replicable by AI.

On the other hand, Product Managers have a different set of responsibilities. They focus on understanding user needs, defining product strategies, prioritizing features, and coordinating with various stakeholders. This role involves strategic decision-making, market research, and collaboration with cross-functional teams. While AI may assist in some aspects of product management, such as data analysis and insights, it is less likely to replace the need for human judgment, leadership, and communication skills.

It’s important to note that the impact of AI on any job role will depend on the specific industry, organization, and technological advancements. Instead of worrying about being replaced, it can be beneficial to focus on developing skills that complement AI technology, such as improving your understanding of AI, data analysis, and leveraging automation tools. Additionally, staying adaptable, embracing lifelong learning, and cultivating a growth mindset can help you navigate potential changes in the job market effectively.

Remember, technological advancements often create new opportunities and transform job roles rather than completely eliminating them.


It’s not about replacing ALL the engineers. What OP is referring to makes sense because let’s say if each Engineer is 50% more productive with AI support, the team now only needs 1/2 the engineering team. It’s not nothing to do with which industry they work in either!


High level languages have made engineers 1000x more productive than programming in assembly. Jobs didn’t decrease.


Therefore, the number of assembly programming jobs today is equal to what it may have been. People who didn’t upgrade to the new skill set didn’t succeed, right?

There will be organisations that perceive this as a boost to productivity, see opportunities to expand in their area, and say wonderful! We now want to produce twice as much.

Some businesses will also delay new hiring or restructuring in an effort to maximize this.

Of course, there is just grey; there is no black and white.


It all depends on how you deploy AI as a tool.

ChatGPT won’t have any idea of how a user is feeling or how they are using your feature, nor will it have any idea of what they genuinely want or need. There will always be a need for that human refinement or touch in frontend.


It can produce front-code from a wireframe based on the language and framework of your choice, but can it do everything? Can it build all the elements, validate, listeners, styles, UI adjustments, and tie in the desired business object? I don’t think it can finish the complete job in a few minutes, either.

Consider it a tool to help you finish your job more quickly and easily. I use it to refine previous writing, get ideas for new topics, and more.


It is capable of producing elements, checks, listeners, unit tests, styles, UI adjustments, and business logic. It can also refine specific aspects of its own code in response to user or stakeholder feedback.


Yes @TerryAnthony, I tested it out on a blazer a few weeks ago, and it worked great. Even if it can handle the fundamental setup, i believe there is still a lot of wiring to be done, especially to get everything precise and relevant to the issue at hand. If you’re really concerned, though, you may gain an advantage and start using it as a tool while also improving your skills to become, say, a full stack developer and ultimately a solutions architect.


Many software engineers lost their careers as a result of compilers because they were no longer required to code in assembly. Interpreters then eliminated even more work by allowing a class of higher compile-at-runtime languages to be used.

You consequently now have a job.

What employment will be available in the future? It’s difficult to say precisely, but they will be present.


@HerbertWarnick Agreed. I’m not saying that software engineers won’t ever find work. I’m making the case that some software engineer skill sets, like programming specifically for the front end, will be supplanted by AI. The importance of other skill sets, such as requirement gathering, will increase. Additionally, new job categories could emerge that operate on an even more abstract level; we are unsure of the nature of these positions. This does not disprove my assertion that fewer engineers will be required overall to complete a given assignment.

1 Like

You do, in my opinion, make a valid point.

Finally, software engineers realise that, as wordsmiths, they are more similar to English majors than they had previously acknowledged. I suppose that’s why it’s referred to as a programming “language.”

It does make sense that AI would be superior than humans at creating code (also known as language).

The fact that the underlying language changes so dramatically for programmers, in my opinion, is the one factor that fundamentally makes this so. It’s not only about using a traditional programming language like C against a contemporary one like JavaScript. It consists of many API versions.

A human is less able than AI to keep up with changing programming languages, standards, and conventions.

The world seems to be waiting for ChatGPT, the no-code utility.