The Rise Of Generative AI and The Future of Software Engineering
In March 2023, Elon Musk signed an open letter asking for a 6-month pause in AI research joining thousands of others including prominent members of the science and tech industry. Ironically, Musk co-founded OpenAI, the company behind ChatGPT, which has propelled AI into the mainstream. Despite signing the petition, Musk continued to set up a new AI company called X.AI Corp. It would seem that Musk acknowledged the potential for generative AI and despite his protestations had also succumbed to the notion that progress can’t be stopped.
The spectacular rise of ChatGPT
To put generative AI into perspective, analysts have compared the uptake of ChatGPT compared to other popular platforms and found it to be the fastest growing application in known history. It took 3.5 years for Netflix to reach 1 million users. It took Twitter 2 years to reach the same milestone. For Facebook, it took 10 months. ChatGPT took just 5 days.
Why the sudden interest in AI when AI technology has already existed for many years, you might ask. Whilst the focus of “traditional” AI has been on processing data and detecting patterns, generative AI goes beyond consumption of existing content to the creation of new content.
Whatever our individual views of generative AI, the technology is already being used in numerous applications across many industries including medical diagnosis, virtual models, video gaming, music generation, filmmaking, and education. You can even use generative AI for more mundane tasks such as creating a slide deck for your next presentation.
Doomsayers will be ever-present whenever new technological concepts near critical mass and I don’t doubt that there will be those who would look to use generative AI for nefarious means, but I get heady just contemplating the limitless possibilities for putting the technology to virtuous application.
The digital concierge
One way in which generative AI adds utility is when it comes to information retrieval. Before AI, we relied upon concepts such as keyword search in order to be able to retrieve information from vast knowledge vaults, but the mere act of searching didn’t always deliver the desired result. Through the careful crafting of search terms, users might be able to uncover relevant bits of information depending on their level of Google-fu, but often with an equal chance of getting back irrelevant noise. The more expansive the underlying dataset, the greater the challenge of getting exactly what you want.
With AI, users are now able to describe the context of the questions for which they seek answers using everyday language until the desired outcome is arrived at. Such are the possibilities that despite dominating the search industry for over 2 decades, Google views Microsoft’s AI-powered Bing search engine as a real threat to its previously untouchable throne of search as consumer electronics giant Samsung was reported to be considering using Bing as the default search engine in place of Google on their devices.
Engineering with AI
That’s not to say that generative AI isn’t infallible. ChatGPT users are constantly reminded that it may generate inaccurate information and many of us will have seen some of the peculiar imagery created by AI prompts.
A major factor in achieving the desired outcome from generative AI is the input it’s given. So much importance has been placed upon the ability to manipulate AI that a new field of “prompt engineering” has arisen with salaries over $300,000.
What does all of this mean for the software engineering industry? Your guess is as good as mine and limited only by our imaginations. ChatGPT is already able to pass AWS certification exams and is being used effectively by engineers in my own team to generate working code in a fraction of the time it would take a seasoned engineer to write. I foresee prompt engineering becoming a staple skill for software engineers.
As an engineering manager, I’m naturally interested in the opportunities and threats posed by generative AI. How could generative AI help my team reduce friction in the ways we currently work? Could we adopt new processes which take advantage of the technology? Are there new possibilities within the software development lifecycle which could benefit from more sophisticated automation? Could generative AI provide us with access to virtual skills from roles we don’t have within our team? What kind of psychological impact does the existence of generative AI have on individuals?
A few examples came to my mind and it would appear that others shared the same curiosity… Imagine services running directly from requirements rather than just code with the ability to self-diagnose when exceptions occur and adapt themselves within pre-defined boundaries. Or how about being able to extract concise answers to questions posed to a digital concierge which has access to your company’s internal wiki. Consider how quickly your company could have reacted to the infamous Log4j vulnerability if it could have quickly garnered information about its entire codebase with a few simple questions.
Conclusion
It’s an exciting time to be in the early days of a technology that has the potential to revolutionise how people engage with technology. Generative AI has the capability to put the power of creativity into the hands of those with merely the means to describe what they desire. No longer will skills or resources act as gatekeepers to previously unattainable goals.
I don’t believe we’re quite ready to make swathes of the workforce redundant. Generative AI is still in its infancy or at the very least at the toddler stage where it can walk and talk, but doesn’t fully understand the nuances of modern society like not being racist or misogynistic. Generative AI still needs a guiding hand so it can develop into a useful contributor and there’s nothing quite like a human to provide a human perspective on things (at least for now).
I think it’s safe to say that there are opportunities to be had and that ignoring the potential impact of generative AI would be akin to how Kodak disregarded the digital camera sensor or Blockbuster discounted the potential of streaming media. “Kodak who? Blockbuster who?” you might ask. Exactly my point.
