How open should AI be?

DANIEL MCCARTHY, SENIOR EDITOR DANIEL.MCCARTHY@PHOTONICS.COM

Google’s strikingly simple search page first appeared the same year this editor first began working for Photonics Spectra. There was, at that time, a healthy number of people who still mistook the predigested content of America Online for the wild west of the worldwide web. More savvy internet users probed the internet’s rich information resources using a collection of search protocols labeled worms, spiders, crawlers, and robots.

Google’s minimalist webpage, in comparison, deliberately made navigating the labyrinthine internet feel effortless. But underlying this simplicity was the highly sophisticated PageRank algorithm that not only revolutionized internet searches but also launched how advertising, marketing, and business were conducted.

Algorithms, computing, and the internet itself have all evolved so rapidly in the ensuing decades that it is often difficult to distinguish hype from legitimate step changes. Remember when voice search was going to change everything?

When OpenAI launched ChatGPT in November, the headlines about the platform’s ability to generate fluent verbal responses to simple queries sounded novel, like a parlor trick. There was a news cycle worth of fretting over what ChatGPT meant to the integrity of high school composition assignments. More unnerving to this office were reports that technology news site CNET secretly experimented with the AI platform to generate published articles. Gratifyingly, the level of errors found in that content temporarily guaranteed the job security of editors, if not writers.

I debated giving ChatGPT the opportunity to contribute a paragraph to this month’s editorial. But the website informed me that the chatbot was busy and would get back to me later. So, we can confirm that it at least talks like an editor.

More recent online discussions about AI chatbots have speculated about their immediate potential impact on developing computer code, generating phishing attempts, and facilitating internet searches. That list prompted an old-fashioned Google search, which unveiled such a volume of other potential uses that it finally became clear that the proliferation of interactive AI could be as disruptive to the status quo as the internet or social media were — and with the same early disregard to potential consequences.

This, obviously, is not a magazine for technophobes. We try, however, to take as unbiased a view as possible on the benefits, challenges, and trade-offs of any given technical solution.

Like integrated photonics, quantum computing, and even laser fusion, AI technology carries the oversize potential to redefine generational cultures, global economies, and the course of human history. It is a tool, a powerful one, that is only as constructive as the people who wield it.
>