DailyPost 1834
CODEX
The battle between man and machine got into its harshest phase with Artificial Intelligence showing its true colors in large numbers of areas impacting our life. There has been the looming fear of AI taking over the technical and domain competence and running the world on its own. Lot of it has been hyped by people who are called the headline creators and headline braggers. The depths and pace of technology and the interconnect for what technology is meant, is beyond their farthest imagination. Lots of data, storage, compute coupled with AI algos meant the death of even the highly skilled technical workforce, was the thought process being propagated. OpenAI has been revolutionizing this field for the benefit of mankind.
The transparency in research and making it available to mankind for its technological advancement is what we owe to them. Codex has just been released by AI company Open AI. ”Codex is a new system that automatically writes software code using simple prompts written in plain language.” GPT-3 the revolutionary deep learning platform is what Codex is based on. It clearly establishes both its pedigree and its future trajectory. It would deliver in the same manner as the GPT-3 and proposed trends already made public for an even more revolutionary GPT-4. GPT-3 has reached crazy levels, it has been ”trained on nearly all publicly available written text on the Internet through 2019.
The most important question being asked the world over, regarding Codex is ”Will this replace human programmers?” Think of the recent Income Tax software fiasco, can a tool of this nature have averted it? The answer is No. This explains the distance to be covered if ever it were to happen. ”The day a non-coder can sit down with Codex, write a spec sheet, and crank out a working piece of software is still far away.” What does a coder actually do and what is the time allocated to different activities, seems like a billion dollar secret that I am letting out. The coder/s, coding team starts the engagement with the client understanding the problem to be solved and the process thereof. It leads to meeting with the individual staff members, stakeholders, agency constituents and more. Ideally, after a few months of work of this nature and having got the specs right, the coding can begin.
Thus the major part of a software project is to understand the problems, “determining the right software and the mathematical models to use, gathering and cleaning the actual data, and presenting results.” In many cases time spent by programmers to write actual code is as little as 20%. AI systems like Codex will do the magic in that 20%. It would make each developer more efficient. It might need lesser coders. But there is a whole world of mid-size and small companies which will provide for downward filtration of software development, ideally suited for them. Talented coders assisted by Codex can change the world. Currently, Codex’s code contains errors or simply does not work 63% of the time. The rational tech thought predicts that, ”Systems like Codex could create “Centaurs,” hybrids of humans and AIs working together to do something faster and better than either could do alone.”
CODEX WILL TURN OUT TO BE THE ULTIMATE TOOL FOR THE CODER, SO TO SAY AN AUTO PILOT.
Sanjay Sahay