start from zero

Still thinking where to get started, so I’ll just go directly to the point. With all the recent development on various models released by OpenAI, Anthropic, Google and other unknown players, it gives endgame vibes. More I think about it, my brain hurts and I get optimistic with all the opportunity, and also get pessimistic with what would happen to current society work dynamics. Let me break it down to you, what I mean - I’ll start with domain which is I am most familiar with - Software development.

Cursor/ Copilot/ Devin/ Bolt/ Marbalism - These are some of the tools I tried in last few months, and the output they generate has pretty much absolutely eliminated many bottlenecks that any software development process encounters for 60-70% of the time. I wouldn’t argue with someone who may question how I came up with that number; let’s just say it’s based on instinct. Historically, all companies have many projects in the pipeline and wishlist. When you approach the development team, their response is to prioritize and determine which project aligns best with our immediate needs, and we will tackle those tasks first. Recent AI advancements have removed those obstacles or on the way to optimize at scale on those, and those timelines have gone from months to days and in some cases just a few hours. The increased context window and advancements has made some tasks dependent on the availability of computing rather than availability of human resources. Now that is a good place to start exploring other sectors. So, let me try to provide more information on those.

Example domains comes to mind -

Insurance Agents and claim processors overall - who verify claim, go through paperwork and provide their recommendation.

Government officials who review permits and find discrepancies or missing information engage in repeated communications with citizens.

Data Analyst (in different domains) who examine reports and provides their recommendation or compile reports.

Legal assistants and paralegals who assist lawyers in preparing their case,

Radiologist who supervises the exams, reads and interprets the images, and writes a report for your healthcare provider (so partial job of just reading and providing a report to some extent)

Curriculum writers - Educational content creators who read the standards and write the appropriate content for age, grade level and standard support.

Sales professionals - In different domain who try to meet where their customers are.

Accounting, Product Managers, Application Testers, Quality Assurance, Research Analyst and many more professions which I am not even aware of and probably don’t understand will have a major impact or are already impacted.

All of these jobs can go magnitude of boost with this blackbox AI models. Without going into too much detail, I can fairly assess that it just works. I am sure people working in all these domains may shake their head in disapproval, but you can bookmark this post and comment back in a couple of years. Keep your receipts and get paid (happy to be wrong, but I don’t think I am in this case.) I don’t think these jobs will be gone, but these job will become exceptionally efficient, meaning it can be done with fewer people compared to what they employ currently and with a lower error rate and much faster. Similar to software development, some of the work will be done immediately instead of months, weeks, and years.

Lot of professions were limited by the amount of analytical, admin, clerical, manual labor (like doing repetitive work like making food - not all types of food, but many other daily ones, driving, etc.) and creative work was limited by the number of human heads involved, which is moving to compute. You can argue that compute is not here yet, but there is no doubt that it will arrive and efficiency from here will continue to increase too. All the things where we argue that but it can’t do A, then models will improve and it will do A, and then you will move to B and argue that but it can’t do B, and it will do B. So you understand, we will assist through crowd work and point out like a skeptic, and someone or some model will improve and do that thing until we push all the limits of this. If you go one level up, what is essentially happening is we are replicating the brain in compute, the difference is, it stays focused on something unlike us humans who have many other responsibilities, distractions, and weaknesses compared to this artificially created brain has one task or more than one task but is absolutely optimized for those tasks and it can perform them tirelessly without complaint and those weaknesses. So we have to redefine our objectives and figure out our own place in the new equation at a higher level.

So few places where I feel enough attention is required are - 1) What the next workforce will look like as many of these jobs may face significant changes. 2) What it means for education sector. We know that education providers don’t adapt quickly for many reasons, not always justifiable, but it’s challenging to conduct thorough analysis and update the curriculum because it takes a long time to assess effectiveness. Though some of the majors are absolutely pointless and we can at least agree on that and help to eliminate those. Another issue is how quickly we integrate AI models and its effectiveness, applications, usage into the curriculum. Just thinking about computer science and related majors, there needs to be an overhaul in what they teach, as those frontend, backend, full stack, system engineering jobs are going to look completely different. Do you really want to teach syntax at this point? Do we need to teach a specific programming language when English is sufficient to interact with these tools and they handle the programming for you?

Enough of ambiguous optimism - pessimism. What sectors can be really interesting Personalization in all sectors is one of the big theme, and agents framework for lot of dynamic and repetitive tasks. You will hear a lot of Agents as we get in 2025, its like how Apps become popular in couple of years of iPhone launch in 2007 and website in 2000s.

Healthcare and Biotech - Your health plans, food plans which were so far limited because of cost, resources, it can be done via this models if they are constantly learning, have access to our health records, your profession information and your goals. It already happens if you are up to date with your annual physical, but following it is hard as it never consistent from either your end or from provider, but more to blame is lot of other obstacles even if you would like to follow. Second major one is, the possibility of personalized medication. Personalized and Precision Medicine is another area where AI is going to make significant impacts in coming years, where we were limited due to technical and research resources, those barriers are decreasing significantly and lot of pharmaceutical companies are already embracing this change.

AI, Machine Learning, Software Development - As mentioned in beginning, some of the tools are already becoming very popular in the past year. On the product development and software engineering, the resource limitations will be close to 0. Like how software provided capability to bring marginal cost of adding more customers to zero and value extraction to infinity with minimal infrastructure cost for most part, imagine if the cost of development itself is coming to zero. I know it sounds superficial, but it’s happening and it will get there. What it allows us to do is create a lot more software everywhere. Even the places which we have not even thought of. So what it means is - We are not limited by the resources available for development or lack of skills but we are limited by the imagination, focus and rigor required to do those things or approach those problems. Not much of an excuse on that part, but mostly we need to break free from skeuomorphism. It is hard, but all the straightforward tasks are already completed.

Creative industries and Robotics - producing more robots who do efficient work, in domains that ensure these systems function properly and we continue automating them. They can be used for security purposes, care provision, and performing difficult tasks in environments that are unsuitable for humans. Design, implementation, and maintenance of these systems.

Sectors where there won't be any impact for the most part - Doctors, Nurses, physical therapist, skilled trades like construction, plumbing, electrician, chefs, firefighters, paramedics and many more of those sectors. Not much of actual impact but those will still improve with support provided by infrastructure and system enhancements.

In summary, AI is here to stay, and the pace of innovation will only accelerate. The challenge isn’t what AI can’t do—it’s how we adapt to what it will do. Whether it’s redefining the workforce, reimagining education, or creating entirely new industries, the opportunity is vast. The question is: are we ready to evolve alongside it? My answer is, we are not ready for the serious enough and it will hurt in short term, but we will come to consensus and adapt as we don’t have other choice.

of course the image is generated with ChatGPT.