Ever since OpenAI’s ChatGPT program created a simple way for the everyday person to widely apply AI technology, the AI hype cycle has risen anew. Time will tell if this is a truly revolutionary time or if things fizzle again, but with Microsoft’s Bing GPT, Google’s Bard AI, and many other players rising up in the space, it’s certainly an exciting time to explore.
Increased productivity at a macro level is the biggest societal promise new AI technology brings. It’s fun to see visual examples, like DALL-E creating AI portraits, but it seems wasteful to automate joys. I’m more interested in low/no-code prototyping, faster and more robust analytics pipelines, and a myriad of dull tasks that we can delegate to free up time for more interesting abstract thinking and frontier-pushing insights.
A simple analogy in the white-collar world is the rise of the calculator and Microsoft Excel. Mental math is still useful to avoid interrupting the flow of off-the-cuff sizing and rapid brainstorming, but the proliferation of pocket calculators from the 1970s onward made basic arithmetic trivial. Nowadays, smartphones, computers, and even browsers have built-in calculator functionality. So, too, does Microsoft Excel, with the added bonus of being able to visualize many calculations in parallel. What might have required reams of paper and calculators, transferring numbers again and again until arriving at a final output table (e.g. reporting a small business’s yearly revenue) can be done manually by 1 person in a nicely-formatted and easily-shareable Excel doc. And so it goes — new companies and technologies can automatically sync various revenues and receipts, leaving one free to export to Excel, visualize it in another medium, or otherwise consolidate and share out when needed.
After playing around with some of the new AI tools, creativity is the main constraint at the moment, not the medium. Just like with web search, you have to learn how to use the tool to arrange the information you need, and it’s worth fact-checking information across reputable media. At the same time, it’s amazing how generative AI can consolidate the world’s information into actionable items. In a few hours, I was able to create a functional updating Rolodex, a test idea I had brainstormed but didn’t have the fluency in Python GUIs to easily execute. ChatGPT couldn’t create a perfectly functional app with all my specifications, but it could get me 80% of the way there (the last 20% relied on my ability to read through and write code myself, as well as learning how to phrase requested tweaks to ChatGPT and move me closer to the solution).
The final thing to remember is the old adage of statistics: garbage in, garbage out. All generative AI relies on the collective work of billions of humans who contribute to our modern “world library”: everything accessible through the Internet, and even more beyond that isn’t. Whatever algorithms we design are limited by experiences we can feed into machines. There is much promise available from AI, but beware those who overpromise. Bias and opinion are inextricably tied to that which we call society, which means they are a part of the data we feed, the programs we write, the questions and decisions we ask machines to make. We can work with the wide spectrum of AI technologies and future tools to dramatically expand our capacity to create, but we should not abdicate responsibility.
Disclaimer: I currently work at LinkedIn, which is owned by Microsoft. Opinions are my own.
Interestingly enough, I wrote this article without leveraging any AI help. I did make the headline photo using AI though (via Craiyon).
Andrew Jabara 