Two neurobiologists from Tel Aviv University, Yossi Yovel and Oded Rechavi, have authored an article in the scientific journal Current Biology, exploring the potential of artificial intelligence to bridge the communication gap between humans and animals.
The first obstacle is the necessity for the language model to employ the animals’ inherent communication signals. This means animals should not be required to learn new signals to interact with humans.
For a long time, scientists have been aware of the signals animals use for purposes like courtship or signaling danger. However, to facilitate AI-mediated communication between humans and animals, the AI must interpret these signals in a wider array of behavioral contexts, marking the second challenge.
The third challenge involves the difficulty machines face in eliciting measurable responses from animals, making it seem as though the interaction is with another species member rather than a machine.
This challenge was notably overcome with honeybees when a group of researchers engineered miniature robots that could mimic the bees’ waggle dance, a method used to communicate the location of food sources to the colony.
These robots successfully directed the bees, although this achievement is limited to a single specific context. The creation of a machine that can inquire about a bee’s desires or feelings remains an unmet goal.
IS IT REALLY POSSIBLE TO COMMUNICATE?
According to two neurobiologists, even after overcoming the three primary challenges, humans may still fall short of fully realizing the desired level of communication with animals.
For example, a future algorithm might be capable of informing humans when a house cat is signaling sadness or affection. Nonetheless, the capability to inquire directly about the cat’s feelings may remain elusive.
At this juncture, the neurobiologists reference a quote by philosopher Ludwig Wittgenstein: “If a lion could talk, we might not understand him.”
“We’ll never manage to ask a cat ‘how it feels’, or explain that ‘ChatGPT’ inadvertently translates to ‘CatGPT’ in French, which is amusing,” Yovel and Rechavi remarked.
This leads to speculation on the feasibility of communicating with primates, who are cognitively more similar to humans.
However, to achieve this, AI models would need to be trained on extensive amounts of primate behavioral data. Chatbots like ChatGPT have been developed using billions of text entries from the internet.
Yovel and Rechavi highlight the fact that such an extensive database of primate behavior simply does not exist.
You may also like this content
- Apple is working on “augmented reality” glasses
- Apple Vision Pro Update: Introducing visionOS 2
- Sony Announces PC Support for PS VR2
-
-
-
-
-
-
-
-
-
-
