AI is changing our world. First, let’s understand what it is.

While poring over a puzzle at the Berkshire Athenaeum in Pittsfield, Mary Giulian-Donohue told me that the concept of artificial intelligence is as remote as ever to her.

“I don’t even know anything about it except it makes my phone even more complicated to use,” Mary said in response to a question.

Sitting a table away, a man chimed in: “I haven’t got the foggiest idea — except that ‘AI’ stands for ‘artificial intelligence.’”

Then, he added: “I prefer human intelligence.”

Others browsing around in the Pittsfield library were more  familiar with AI.

“It’s interesting how you can get the computer to actually talk back to you after you ask it a question,” said Russ Lehmann, a retiree. “It seems to be pretty smart these days.”

But while some members of the Berkshire community might not know everything about AI, more people are using it than ever before.

A Pew Research Center survey from June reported nearly half of U.S. adults use AI chatbots, such as ChatGPT, with one in four of them using it every day. About a third of American homes have devices with AI, such as a doorbell, thermostat or smart speaker.

While use is only growing, the Pew study also came with a “but.”

As in, “But Americans — including younger adults— are deeply skeptical of AI. More adults predict that AI will have a negative rather than positive impact on them and on society. And majorities think AI is advancing too quickly and will put their personal information at risk.”

First, what is AI?

And so the question becomes: How is AI actually going to affect Berkshire County in the coming days, months, years? And how are the Berkshires going to respond to it? 

But before doing that, what is AI?

Here’s one answer: “Artificial intelligence is the ability of computers or machines to mimic human intelligence by learning, thinking, solving problems and making decisions.”

That definition was, in fact, given by ChatGPT, a artificial intelligence tool developed by a company called OpenAI.

But was ChatGPT, without the human touch, missing anything in its answer?

Not according to Gerol C. Petruzella, the academic technology consultant at Williams College and a scholar who holds a doctorate in philosophy. Petruzella’s “approach to technology is informed by human values,” and that puts him squarely where people and tech intersect.

“I would define AI as any technology that does things that traditionally would have required human cognition to do,” he said.

AI has a history

The truth is that AI, based on the above explanation, isn’t very new.

The term was coined at a Dartmouth College conference in 1956. Stanley Kubrick’s “2001: A Space Odyssey” grappled with AI on the big screen in the 1960s.

So, if artificial intelligence, as a concept, is old news, what about it has changed to make it new news today?

Today’s artificial intelligence is generative. 

Generative artificial intelligence, or generative AI, is “a type of AI that can create new content — including text, images, audio, video and computer code” — “in response to user prompts.” (Thank you, again, ChatGPT, and Google AI Overview as well.)

Generative AI learns by devouring an internet full of data.

How is this different? Traditional AI technologies, like the iPhone’s Siri and older computer models, used limited amounts of data and worked within preexisting code to help users solve problems.

AI enters the age of creation

Now?

AI technologies, such as ChatGPT, are all about creation.

Generative AI relies on “astonishing quantities of data” and “astonishing amounts of computation power” to make new things, said Petruzella.

This is why AI leaders are insisting on building massive data centers across the United States to give generative AI the physical space it needs to work at its peak.

What generative AI can do is impressive. Take the large language model, or LLM, for instance.

“It’s in the name, right? Those are models that are specifically designed to output language,” Petruzella said.

In other words, a large language model can analyze massive amounts of data in the blink of an eye and use that information to talk like a human would. The words being produced by a large language model are all original.

These large language models include OpenAI’s ChatGPT, Google’s Gemini and Anthropic’s Claude, each of which have a wide range of abilities. Students are spending more time with these tools in schools, and businesses are relying on them more to get work done quickly.

AI changed almost overnight

Large language models aren’t the only examples of generative AI, said Petruzella.

“Generative AI doesn’t have to be about language, right?” Petruzella said. “There are fantastic generative AI models that are trained on biological data, on chemistry, and on architecture.”

All of this growth in this level of artificial intelligence has happened in a flash.

In 2017, Google engineers published about the “Transformer” framework, which opened the floodgates and made it possible for artificials intelligence models to use more data at once than ever before. By late 2022, OpenAI unveiled GPT-3, or the Generative Pre-trained Transformer 3, to the public. Afterward, The Atlantic declared the college essay dead.

Since then, competing companies — Google, OpenAI, Anthropic and more — have been racing to make increasingly advanced generative AI models.

“It’s incredibly fast-moving,” Petruzella said. “The technology has tended to change over the last handful of years, sometimes almost on a monthly basis.”

AI moves from answering to acting

Many of these AI models are described now as “agents,” able to deliver information to people on a regular basis without being prompted by humans.

Thus, “agentic AI” is “empowered to take actions on its own initiative or at least in the absence of a specific human starter,” Petruzella said.

For example, someone can ask an AI agent to sum up unread emails or give a daily news briefing, and the agent will then begin to do this on its own, like an unpaid personal assistant.

AI has reached nearly everybody who uses a phone.

“It’s hard to think of the type of digital tool or platform or app that is not currently integrating artificial intelligence features to some extent,” Petruzella said. “It’s coming to be part of the environment that we work and live in.”

And while AI offers new potential, it also poses a challenge, according to Petruzella.

“Our thinking from this point forward is going to have to take into account that humans are not the only entities anymore that can take actions on their own,” he said. “We have to figure out what the rules of the road are now.”