The pace of the adoption of generative AI (ChatGPT is one implementation of genAI) continues to be extraordinarily rapid. If anything, genAI’s adoption and fears of the consequences of its adoption are accelerating, which is as interesting to me as the technology itself and a determinant in its perceived threats.
Our fears over genAI are nothing new. Most existed long before the neighbor’s cat caught its first rat.
Last week, technology executives gathered at the Capitol to discuss regulating genAI with lawmakers. Mark me down as a skeptic: on the technology executive side, the participants were jockeying for advantage in exploiting the technology. They would have passed if they weren’t slathering to exploit genAI. They argued for restraining their competitors. But themselves? Oh no.
The lawmakers were striking poses against the evil tech execs and trying to establish genAI creds with their constituents. Little was accomplished, although, predictably, everyone promised to continue the discussion. At least while the cameras roll and the press clamors for coverage.
If the lawmakers are serious, which they very well should be, I suggest they call in genAI engineers, scientists, and academics. The discussion would take more effort to understand and generate fewer sound bites. The proceedings would look more like a classroom than a carnival. Far less idle entertainment, but not a complete waste of the time and resources.
Our fears over genAI are nothing new. Most existed long before the neighbor’s cat caught its first rat.
We fear AI will develop an unstoppable super virus that will kill off humankind. Moses used that threat to goad Pharoah into listening to the brickmakers local. But what stopped the Covid-19 pandemic? Vaccines, not tracking down the origin of the virus.
Maybe AI-aided hackers will take over the power grid. Plain old hackers did quite well at shutting down Iran’s uranium purification plant, as did the creeps who brought down Saudi oil refineries. Better and stricter cybersecurity would have stopped both those efforts. AI had nothing to do with it.
What if cheating with AI vitiates education? Cheating didn’t start with ChatGPT, nor will it end if genAI disappears. Students who know the value of learning, not the illusory advantages of arbitrary ticket punches, don’t cheat. Convince students that they will be rewarded in life for what they learn, not the grades they receive, and cheating will be gone.
Which brings us to the biggest fear of all: loss of jobs.
The disappearance of typing pools shows how genAI will change the job market, although crystal ball weather is always cloudy.
Desktop computers and copy machines obliterated typing and stenographic pools starting in the 1970s. Eliminating those pools transformed the nature of office work, and the role of women in the workplace changed dramatically. Women in the office today are far more numerous and significant than they were in the 40s, 50s, and 60s when most were typists, stenographers, and secretaries.
Did the replacement of typing and stenography with copy machines and desktop computers drag women kicking and screaming to become technicians, managers, and executives? Or did the desire of women for greater agency inspire machines that replaced typists and stenographers? You may prefer chickens or eggs but we’re still talking poultry. The workplace of Mad Men is gone.
The disappearance of typing pools shows how genAI will change the job market, although crystal ball weather is always cloudy.
Some kinds of desk work will change radically. The ubiquity of computer network based communications (the internet and the world wide web) has flooded us with words. Has the inundation improved or degraded written discourse? Opinions differ. There’s good writing on the internet, bad writing, and, predictably, a ton of mediocre writing, good enough to convey its intended message, but not of much merit in itself.
Starting in about 2000, some people have made a living producing mediocre network content. They write coherent and passable paragraphs about any subject, not unlike typists and stenographers who transformed anything dictated to them into words on paper. Some of these folks are being replaced by ChatGPT. Since using tools like ChatGPT is cheap and easy, more will be replaced.
GenAI has flaws. It makes things up (the current term is “hallucinates”) unpredictably. Its output is often boring and lifeless, sometimes nonsense. GenAI recipes range in quality from average to inedible. You can bet that developers are working nights and weekends to address these and other genAI issues.
In the 1980s, word processors froze periodically. It’s rare now, but they still do. Bugs still crawl through their algorithms, but only a tiny fraction compared to early times. I put in a few nights and weekends myself killing word processor bugs. They still aren’t perfect, but the easiest place to find a typewriter now is a museum, not an office.
In the 1960s and 70s, most women quit calling typing a career. They were no longer only trained fingers that operated a machine for putting words on paper in conventionally accepted spelling.
What will those who lose their jobs to genAI do and how will work change with genAI? I predict more jobs and more words. Some will move on to jobs they prefer like dog walking or nuclear physics, others will muddle on doing whatever comes next, but the flood of words will not abate.
Perhaps, faced with competition from mediocre genAI, the general quality of internet writing will improve. Yeah. Right.