What Surprised me About AI

According to a statement published by the Washington Post, the newspaper is admitting that it used an AI reporter to publish some 850 articles within the previous year. 

While that sounds like a major admission, you aren’t going to see any headlines about that statement. It’s not because it isn’t newsworthy, but because they made it back in September 2017.

At that time, the Washington Post was talking about its “robot reporter”, named Heliograf, which it had used to generate hundreds of stories. 

According to the Washington Post, Heliograf was primarily used for data-driven stories that, under normal circumstances, wouldn’t receive coverage all. This includes high school sporting events, company earning reports and smaller election results.

The story didn’t make headlines at the time. It was seen, if anything, as a curiosity. However, by early 2019, concerns about “The rise of the robot reporter” were growing as Forbes, Bloomberg, Reuters, the Associated Press and The Guardian all acknowledged increased usage of machine learning in the newsroom.

For anyone paying attention, it was clear that generative AI writing was on the horizon.

In that context, the sudden rise of AI isn’t a surprise at all. We’d known about the technology for over half a decade, and had seen other indicators well before that. It was not a shock that generative AI writing made its debut. 

What surprised me, as well as many others, is the way it made that debut. 

AI followed a path that almost no other technology had and, in doing so, it created unparalleled levels of uncertainty and excitement at the same time.

The Sudden Debut

On November 30, 2022, the AI development company OpenAI officially launched ChatGPT, their AI bot built on GPT-3.5 and GPT-4 large language models. 

The launch was immediately available to the public and for free. Anyone could sign up for an account, engage with the ChatGPT prompt and receive generated text in reply.

According to a report by Jeremy Kahn at Fortune, the launch was something of a last resort for the company, which had been struggling under financial pressure. Even the executives at OpenAI were surprised at the rapid growth of ChatGPT. 

Regardless of the reasons for the launch, it represented a massive and immediate shift in who could access AI-generated text. Previously, generating any significant amount of AI text required either a great deal of money or technical expertise. ChatGPT was both free and could be easily accessed just by creating an account.

Because of that, ChatGPT exploded onto the scene. It obtained more than 1 million users in just five days and reached over 100 million users in less than two months.

That explosive growth did not go unnoticed. Google and Meta both accelerated their own AI development. Microsoft, for their part, extended their partnership with OpenAI

AI, rightly or wrongly, became the buzzword of the entire tech community and its influence can be found almost everywhere. From concerns over AI plagiarism in the classroom to one of the sticking points prompting the WGA strike, AI is wielding tremendous influence over almost all areas of creativity.

However, this wasn’t the debut that most expected. While most had known that this technology was likely coming, not many expected it to come all at once.

Internet Access and Self-Driving Cars

In general, complicated technology, especially technologies with potential societal implications, are generally released gradually, giving at least some time for the public to come to grips with the tech and the implications it may have.

An interesting contrast is self-driving cars. Though Tesla first introduced its Autopilot in October 2015, some elements of the technology, including lane assist, were already in vehicles by the mid-2000s.

However, even the debut of Autopilot wasn’t a coup for the technology. Limited only to a subset of Tesla vehicles, it wasn’t available to other brands, nor could older vehicles use it.

Since then, the sheen has begun to come off of Autopilot, Tesla is facing both a lawsuit filed by its own drivers and a criminal probe into allegations that it fails to meet its self-driving claims

Most drivers, even eight years after the debut of Autopilot, have still never driven nor even ridden in a full-self-driving car.

However, what many more drivers have had access to is companion features such as the aforementioned lane assist, assisted braking, automated parallel parking and more. These features are a combination of safety and comfort improvements.

It seemed likely that Generative AI would have followed a similar pattern. Limiting use of the actual technology to a small subset of the population, but allowing others to enjoy comfort benefits such as improved grammar checking, AI editing of human works and other writing improvement tools.

Instead, it’s almost as if every car, truck and motorcycle were provided access to a full self-driving mode overnight, regardless of how good that mode is, how safe it actually drives or how it might impact others on the road.

Of course, that was never possible with full self-driving, in addition to regulations that would prevent it, equipping older vehicles with the technology would be a mammoth task that would, most likely, take decades to perform on a massive scale. All OpenAI had to do was, quite literally, flip the switch.

There were many reasons to think neither they nor their competitors would do that. Namely, AI generation is still resource intensive and the cost of development has been very high. The decision to give it away to the public for free was certainly a surprise to many, even those who were closely watching the space.

Because of that sudden release, AI didn’t get much of a transition phase, at least not a public-facing one. That, in turn, is why there is so much struggle to keep up with the changes it’s bringing, let alone either fight against those changes or support them.

Bottom Line

To be clear, it’s still very uncertain what long-term impact AI will actually have in its current form. It could turn out to be like cryptocurrency and NFTs, a huge fad that the world got excited about, but ultimately only really mattered to a small group of people. 

There are signs that could be happening. Trust in ChatGPT is already beginning to waiver in business circles, and headlines about the bot’s poor performance are now incredibly common.

It could also wind up being a complete game-changer that alters our ideas of authorship, productivity and more. It could wind up being the most important tech advancement in decades. 

The likely truth is somewhere between those two extremes. Still, one thing is clear, for the near future, it’s likely that AI’s role will be shaped almost as much by how it debuted rather than what it is actually capable of. 

AI definitely made a tremendous splash, but that splash is representative of how it entered the water, not how much weight or impact it will have on the pond. That’s something that will still be born out over time. 

Want to Reuse or Republish this Content?

If you want to feature this article in your site, classroom or elsewhere, just let us know! We usually grant permission within 24 hours.

Click Here to Get Permission for Free