Birds and AI May Have a Lot in Common—Here's Why That's Reassuring

Artificial intelligence (AI) is taking over everything from writing to drug discovery, but this form of computing may not be as smart as it appears. 

A new study examined the workings of the pigeon brain and how the bird's learning shares similarities with artificial intelligence. Experts say studies like this show AI general intelligence is still far from reaching human levels.

"Most AI today is still narrow or weak AI, meaning it has been developed to solve specific problems or support a narrow range of activities," John Benamati, the chair of Information Systems and Analytics at the Farmer School of Business at Miami University, told Lifewire in an email interview. "Speech interpretation and text recognition that has been able to interpret and respond to words are not new. These types of AI, such as Alexa and SIRI, have been around for a while. Recent advancements have enhanced AI algorithms to appear even more intelligent."

Limits to AI General Intelligence

A pigeon study at the University of Iowa had parallels with computing. Scientists administered complex categorization tests that the birds couldn't solve using high-level thinking. Using trial and error, the pigeons memorized enough scenarios in the test to reach nearly 70% accuracy. 

"You hear all the time about the wonders of AI, all the amazing things that it can do," Ed Wasserman, a professor of Experimental Psychology at Iowa University and the study's corresponding author, said in a news release. "It can beat the pants off people playing chess, or at any video game, for that matter. It can beat us at all kinds of things. How does it do it? Is it smart? No, it's using the same system or an equivalent system to what the pigeon is using here."

Benamati said recent advances had enhanced AI algorithms to appear even more intelligent than they actually are. 

"Generative AI such as ChatGPT and online intelligent agents such as Google's Duplex can now more accurately understand both the 'words' and the context in which the words are communicated," he added. 

David Ly, the CEO of Iveda, a company that makes cloud-based AI video and sensor technologies, told Lifewire via email that AI is currently limited by its inability to understand context and emotions in the same way humans do. 

"AI systems rely on large amounts of data to make predictions and decisions, and they are only as good as the data they are trained on—meaning, AI can produce biased outcomes if the training data is not representative of the real world," Ly added. "This is a growing concern in the AI community, and researchers are working on ways to ensure that AI systems are fair and unbiased."

When Will General AI Happen?

While AI is far from matching human intelligence, some observers say computers may eventually catch up. 

"Is it possible for AI to eventually overtake human intelligence?" Benamati asked. "I think the answer is yes. Is it probable? I am less convinced, but at the pace of advancement, I think we should be considering it more seriously than we seem to. I think we are still far from general intelligence in AI, but not nearly as far as we used to be."

Touro University Graduate School of Technology professor Jeremy Rambarran said in an email that he doesn't think fears about AI Intelligence overtaking human intelligence are overblown. 

"In fact, these fears can be justified as bots can help enhance some human-related tasks by intervening and doing constant checks and then allowing humans to perform other tasks that cannot be done by simply using a bot," he added. "However, depending on its application, bots that are heavily infused with AI can reduce the need or requirement for human workers."

However, Rambarran suggested that we are decades away from creating general AI, a term that is often defined as the ability of an intelligent agent to understand or learn any intellectual task that a human being can. 

"Perhaps in the nearest century, when technology becomes more advanced, we can have a general AI Intelligence or at least a full algorithm, which can be applied to any or all types of applications and bots," he added.