I tried ChatGPT and it is quite amazing. I could do things like (for example) write a program to calculate square root of a number or draft an article on Anti-microbial Resistance in humans - both subjects I know nothing about!
Of course, ChatGPT still cannot do many other things, but this is just the beginning, Difficult to imagine where humanity is headed.
I agree, but from what I have seen till now, all automations in turn help improve developer/analyst’s productivity.
To give an example, there are workflows which automates data extraction, data operations, update the visualization dashboard (Tableau, PowerBI), attach the file link and share this in an email to the specified recipients in just a few clicks (ideally just a single click)
ChatGPT, if leveraged in the right way would help minimize intellectual effort in manual tasks and help humans spend time on better, more creative things.
Yes, we cannot ignore the risk of reduced IT labor which time will tell. Usually automation paves way for better tasks.
The chatbot doesn’t just do tasks (that’s automation), it can program. Ability to write code is a higher level of intelligence. If machines can code, do we need I.T. companies? Clients can do it themselves. Or at least, what is today a 100-man-day job can become a 10-man-day job. Presently most of the project pricing is decided based on “effort estimates” (billing clients on how much time it takes to write X lines of code). But what happens if much of the code is available off-the-shelf, how will the company justify its pricing - that is the question.
For a set of possible scenarios where every possible scenario either which has happened or thought of is created, solutions can be made with the help of a code, as this is absolute quantification, only numbers are changed. But when each claim is different, something like in an insurance sector, where the applicant can come up with his personal agenda which cannot be thought of and which is made just because it is his legal right and he has a chance of winning, a prior coded model does not work. Randomness cannot be modeled.
Also, tech in itself is wonderful but the creators of that tech may not possess the necessary knowledge to value it, so the behemoths of IT with their cash flows and reserves always have the chance to buy out such tech, along with their in-house R&D.
So I guess as the sector grows, so does the established companies, unless they make grave miscalculations in assessing something, which make them become small.
Client can do it themselves even now. What a chatbot do can be done by an intern or by dime a dozen freelance programmer from sites like Fiverr as well. Companies don’t want just a working code but a code which is fool proof, robust and secure. There are banks which are still running on 90s mainframe software as they don’t want to interrupt tested software. It is still a long way for companies to rely on AI to churn new code. In fact they might need these service companies to try all these. Or better, the service companies themselves will make use of these systems to bring in efficiency and reduce bench strength making them more productive per employee.
Coding as an activity is hardly 1/3rd (actually less than that) of the overall work in any IT project.
Code is written in a business context, business context exists within a larger capability, the capability is at some lifecycle stage, the industry dynamics, competition, and so on. One can automate the bottom layers but it ain’t easy as you navigate the upper layers.
Tech can/will help in productivity, and supplement decision-making but not replace humans for non-repetitive tasks. Case in point - self-driving cars. Despite all the hype for more than a decade now, here is what the flag-bearer Tesla said recently.
Service companies need to embrace these realities. It must be seen as an opportunity than to be threatened.
Every tool that is in existence since decades writes code in background for every customisation. Code is nothing but logic in a different language. These tools also need IT companies for integration and consulting to meet business requirements and customisation…so tech may change, way of coding may change, those who do coding may change but an evolving company/industry may not be directly threatened by it is what I feel…
@Chandragupta sir, very apt example and yes the ability to program at this point of time would cater to certain problem statement, the only caveat is, applications are developed in agile methodology where with iteration, the increments/feature addition/ui enhancements are done based on business context.
For this to be be done by chatgpt, the client would need to describe the problem statement very specifically instruction wise, basically translate a real world case into something a machine or “AI” would understand, this itself is a niche skill for someone to know both things.
Yes, in future there would a very enhanced version of ChatGPT which might even do that.
Reminds of Robot movie where despite the robot being intelligent, doesn’t always understand the context in the way the master would want let’s hope for the best.
If AI can write code, human programmers would still have their job. They would be the ones who ask the AI in the right way about what to write code for. We have to remember that writing code is only one part of programming.
Currently, programmers write in computer language to get stuff done by computers. If AI writes in computer language, programmers will write in human language to interact with the programming AI to get stuff done by computers.
Conservatively, AI would become another tool in the toolkit of programmers. There is no question of replacing programmers. Their work will just become a bit easier to do certain parts of their work.