Just after hatching, many birds learn to identify and follow the first moving object they encounter—a process called imprinting, which can offer protection in the wild as it helps them stay near a parent. It doesn’t take much visual information for a bird to learn to prefer one object and follow it. Researchers wanted to know whether AI models called transformers could do a similar task with limited inputs.

Transformers are generic learning systems that can be trained to perform a wide variety of tasks, making them useful in both AI chatbots such as ChatGPT and in computer vision applications, such as autonomous car navigation.

“To directly compare learning algorithms (计算程序) to brains, we need to train them on the same experiences,” says Samantha Wood at Indiana University Bloomington. She first raised chicks in a box where the only visual stimulation came from a rotating 3D object presented on a screen. After the first week, she ran each chick through hundreds of test trials that showed that same object on one screen－presented from both familiar and unfamiliar perspectives－and displayed a second unfamiliar object on another screen. The chicks spent more of their time near the first object, suggesting they had imprinted on it.

The researcher then created a virtual simulation (仿造物) of the set-up and used a virtual agent to move through it while looking around and recording a first-person view. That provided tens of thousands of simulated images for training and evaluating four transformer models.

The AI models had just 300 milliseconds to learn from each simulated image－approximating (接近于) how long biological neurons (神经元) fire after being presented with an image. The researcher found that the AIs could learn to recognise a 3D object as quickly and accurately as the chicks.

The study is “a great piece of work” in comparing machine performance with biological brains, says Antone Martinho-Truswell at the University of Sydney. But he also notes, “We might be able to say that the chick ‘saw’its imprinting object, but that will have a component (成分) of experience to it. Particularly as imprinting is to do with identifying its mother, it would be unsurprising if that visual experience were combined with a suite of other components of experience: fear yielding to comfort, for example, as the chick comes to regard the object as its imprinted ‘mother’.”

【小题1】Why do newborn birds engage in imprinting?

A．To enhance their navigation skills.

B．To develop their social behaviour.

C．To improve their communication with other birds.

D．To establish a protective connection with a guardian.

【小题2】How did Samantha Wood conduct the initial experiment with chicks?

A．She raised them in an environment with a rotating visual element.

B．She exposed them to various visual stimulations in the wild.

C．She showed them various moving objects on screens.

D．She observed their behaviour in a natural habitat.

【小题3】What role did the virtual simulation play in the research?

A．To imitate the natural behaviour of birds.

B．To assess the effectiveness of virtual agents.

C．To examine the Al models’ability to identify a 3D object.

D．To create a visually diverse environment for the chicks.

【小题4】According to Antone Martinho-Truswell, what poses a challenge in comparing machine performance with biological brains?

A．Rapid learning pace of AI models.

B．Recreating real-world environments for experiments.

C．The complexity and diversity of biological experiences.

D．Conducting additional experiments with a range of animals.