The lesson about AI that a precursor to ChatGPT left us in 1966

“I am existing.”

That's how Cortana, Microsoft's virtual assistant, was instructed to respond in 2014 when asked if she was alive.

Let's go forward in time, and today, we face social and philosophical challenges brought by artificial intelligence technologies, such as the ChatGPT, which is part of the Bing search engine.

While these technologies have advanced abilities, such as the ability to collaborate, they also raise concerns about increasing human involvement with machines.

However, history indicates that it is unlikely that we can significantly curb its development. By looking at past interactions between people and robots, it is more likely that in the future we will accept and even adapt to them as a “family”.

This may have consequences that we are not yet able to foresee.

Eliza – the first virtual assistant

Virtual assistant technology dates back to the 1960s, when computer scientist Joseph Weizenbaum at MIT designed Eliza. This natural language processing program could convincingly imitate short human conversations, and in one famous application, it could simulate the interaction between a client and a therapist.

Photo: Google.

Although it ran on scripts and pattern matching, users were impressed with its ability to mimic human conversations. The program has been experienced by students and colleagues, including Dr. Sherry Turkle, who has since studied the social effects of machines.

Despite being designed as a parody of the doctor-patient relationship, users spoke to Eliza, attributing her intelligence and compassion. Although Eliza's creator made it clear that the program lacked these capabilities, he was convincing enough. enough that Weizenbaum's secretary asked him to leave the room so he could speak with Eliza in particular.

It wasn't until the 2010s that virtual assistants like Siri, Cortana and Alexa achieved widespread adoption, but the history of technology's forerunners began with Eliza more than fifty years ago.

In its practical application, Eliza was constrained and not intuitive, requiring the programming of new interaction patterns. However, the tendency of users to attribute realistic abilities to Eliza was an important finding, contradicting what the creator, Joseph Weizenbaum, hoped to show.

As he later wrote, he failed to realize that short exposures to a simple computer program could lead normal people to delusional thinking.

Humans tend to humanize machines

Dr. Sherry Turkle explains that this tendency of humans to attribute emotions, intelligence and even consciousness to machines is known as the eliza effect. It's a result of our tendency to create robots in our own image, connect with them easily, and make ourselves vulnerable to the emotional power of that connection.

In short, users brought life and personality to a rudimentary chatbot that had no learning or generation capabilities.

People tend to push the boundaries of virtual assistant designs and seek interactions for the which they weren't designed, including declaring your love, proposing marriage or talking about your days. These human needs create the foundation for relationships with chatbots, which, thanks to advances in machine learning, feel more spontaneous and social than their predecessors.

Jake Rossen writes of Eliza's reception and comments that in the 1960s it was a tempting flirtation with machine intelligence, but its creator, Joseph Weizenbaum, was not prepared for the consequences. Now, as we enter a time in history where virtual assistants are increasingly common and accessible, we are still unprepared for the consequences not only of their abilities but also of our tendency to welcome and accommodate them, sometimes to the detriment of own.

Source: salon