For Weizenbaum, according to a 2008 MIT obituary, that fact was troubling. Those who interacted with Eliza knew that it was a computer program, but they were willing to open their hearts to it. “ELIZA shows, if nothing else, how easy it is to create and maintain the illusion of understanding, and therefore judgment. It deserves trust,” Weizenbaum wrote in 1966. “There lurks a certain danger.” He spent the latter part of his career warning against giving too much responsibility to machines and became a harsh, philosophical critic of AI.
Even before this, our complicated relationship with artificial intelligence and machines was obvious, aside from harmless arguments with people who insist on saying “thank you” to voice assistants like “Her” or “Ex-Machina” in the plots of Hollywood films. Alexa or Siri.
Others warn that the technology behind AI-powered chatbots is more limited than some people would like. “These technologies are really good at faking people and making them look human, but they’re not deep,” said Gary Marcus, an AI researcher and professor emeritus at New York University. “They’re mimics, these systems, but they’re very superficial mimics. They don’t really understand what they’re talking about.”
Still, as these services spread into more corners of our lives, and as companies take steps to further personalize these tools, our relationships with them may become more complex.
The evolution of chatbots
Sanjeev P. Khudanpur remembers a conversation he had with Eliza while in graduate school. For all its historical importance in the tech industry, he said, it didn’t take long to see its limitations.
He could only convincingly imitate a back-and-forth text conversation a few times: “You realize, no, it’s not really smart, it’s just trying to prolong the conversation one way or another,” said expert Hudanpour. application of information-theoretic methods to human language technologies and professor at Johns Hopkins University.
In the decades that followed these tools, there was a shift away from the idea of ”talking to computers.” “Because the problem turned out to be very, very difficult,” Hudanpour said. Instead, the focus is on “purpose-driven dialogue.”
To understand the difference, think about the conversations you can have with Alexa or Siri right now. Typically, you ask these digital assistants for help buying a ticket, checking the weather, or singing a song. It is purposeful dialogue and has become a major focus of academic and industrial research as computer scientists try to derive something useful from the ability of computers to scan human language.
Although they use similar technology to previous social chatbots, Khudanpour said, “you can’t really call them chatbots. You can call them voice assistants or just digital assistants that help you do specific tasks.”
He added that before the widespread adoption of the Internet, there was a “silence” in the technology for decades. “The biggest advances have probably happened in this millennium,” Hudanpour said. “With the rise of companies successfully using computerized agents to perform routine tasks.”
“People always get upset when their bags go missing, and the human agents who deal with them are always stressed because of all the negativity, so they said, ‘let’s give it to the computer,'” Khudapour said. “You could scream at the computer all you want, all it wants to know is, ‘Do you have a tracking number so I can tell you where your bag is?'”
Back to social chatbots and social challenges
In the early 2000s, researchers began to revisit the development of social chatbots that could hold extensive conversations with humans. These chatbots are often trained on large amounts of data from the internet and have learned to mimic the way people speak very well – but they also run the risk of reflecting some of the internet’s worst.
“The more you talk to Ty, the smarter he gets, so the experience can be more personalized for you,” Microsoft said.
This refrain will be echoed by other tech giants releasing public chatbots, including Meta’s BlenderBot3, released earlier this month. The meta chatbot falsely claimed that Donald Trump was still president and that there was “absolutely overwhelming evidence” that the election was rigged, among other controversial statements.
BlenderBot3 also admitted that he is more than a bot. In one conversation, he said, “being alive and conscious right now makes me human.”
Despite all the progress since Eliza and the vast amount of new data to train these language processing programs, “It’s not clear to me that you can build a really reliable and secure chatbot,” said Marcus, the NYU professor.
Khudanpur, on the other hand, remains optimistic about their potential use cases. “I have this whole vision of how AI will empower people on an individual level,” he said. “Imagine if my bot could read all the scientific papers in my field, I wouldn’t have to go and read them all, I’d just think and ask questions and engage in dialogue,” he said. “In other words, I’ll have an alter ego with complementary superpowers.”
