I’m sure you feel constantly challenged by technology changing all the time. As a software development professional, I feel challenged, too. Take the technology company OpenAI and its innovations on generative AI. The company introduced ChatGPT in November, 2022 reached 100 million users in three months and released a new version in mid-March—a new bar in exponential growth.
Though using tools such as ChatGPT to write software is still under discussion, there is no question that coders have made specific tasks within the software development process more efficient by using AI as a productivity tool, improving the way they work.
AI in action
Software developers have recently started using AI for things like gathering requirements, analyzing business opportunities and designing new architectures. During the implementation phase, they are using AI to jumpstart code writing, review new code and create datasets to speed up testing. Finally, AI-improved performance monitoring tools are helping make deployment and maintenance more effective and precise.
Think about that impact in one of the industries with significant volumes of demand for software development: financial services. Imagine what we could accomplish if we were to use AI to update the underlying software, much of which is decades old, of all financial transactions more quickly in the United States.
Today this feels like a dream because updating banking systems is a tough challenge. It takes years of work and such massive investments that only the most prominent financial institutions in the country can barely afford it.
The essential G in GPT
This is no exaggeration: the CEO of Truist said at the time that only the merger behind the new bank, the sixth largest in the United States by deposits, in December 2019 would give them the scale to afford all the technology they needed.
Who is going to update the software of the 4,000 banks in America smaller than Truist, which are also being challenged and disrupted by fintech companies and digital banks? Who can afford to write new software for all the acquirers, processors and merchant services providers of these traditional organizations?
The answer is that we will be able to afford all the new software we need because the new generative artificial intelligence software, the “G” in GPT, is highly efficient at helping bright talent disrupt how we write new software.
How efficient? Last March, OpenAI President Greg Brockman showed his new system a handmade drawing of a website in his notebook and returned a functional website with working code in seconds. And his handwriting is atrocious.
Productivity multiplied
At Compass UOL we looked into how much of our work we could automate in March 2023, and we learned from our software engineers, customers and our partners that they have already achieved a 38 percent productivity increase, on average, by using AI tools in the last five years.
Our researchers predicted that by 2025, we will achieve a 114 percent productivity gain, using the same benchmark. The software development tasks to solve a specific use case within the scope of Compass’s study took an average of 78 hours a few years ago before AI was introduced in the process, took 56 hours with the AI we employed through 2022, and are estimated to take 36 hours with new Gen AI tools by 2025.
That is less than half the time writing software took everyone in the industry around 2018, not that long ago. It will be a huge shift in how we build digital platforms.
This doesn’t mean that software developers are no longer required: they are needed more than ever. It is just that we no longer need to start from scratch. We will automate the things that take the longest and that we enjoy the least—like writing manuals and creating use cases—and will spend more time on creative and complex business discussions to build the best applications that allow leading companies from different industries to thrive.
