What does he have better than [Pompon]?
Rien(nothing), answers Aurelie in a shaky voice, barely above a whisper.
Its not clear whether shes talking about the stray alley cat or the shepherd boy.
It’s free, every week, in your inbox.
You sayrien, Aimable tells his wife in a sweet and soft voice.
Again, there are many hidden meanings and accusations in his words.
And thats where he drives the dagger.
Look there, he says.
This is why she returned.
She was cold and hungry.
Meanwhile, Aurelie is holding the heart-shaped bread that Aimable had bakedfor himself.
We can empathize with Aimable and Aurelie (and map them to our own previous life experiences).
But, Mitchell suspects, something more fundamental might be missing.
They also examined to what extent current AI systems enjoy these capacities.
Understanding is built on a foundation of innate core knowledge, Mitchell writes.
Neural networks cant extrapolate
Compared to humans,deep neural networksneedmuch more data to learn new things.
Abstraction is a powerful tool of the human mind.
The AI and neuroscience community is divided on how the human mind acquires knowledge efficiently.
Many scientists believe thatthe brain comes prewired with many capabilities.
For instance, while chimpanzees are obviously less intelligent than humans, they have a much better short-term memory.
Likewise, the minds of squirrels have evolved to remember thousands of food hideouts.
These are cognitive abilities that have developed over thousands and millions of generations and repeated interactions with the environment.
Theres no single metric to measure the level of understanding.
Another problem that plagues the AI community is the narrow focus on optimizing algorithms for specific benchmarks and datasets.
These datasets allow AI researchers to train their algorithms and test their accuracy and performance.
When scoring higher on dataset becomes the goal, it can lead to detrimental results.
The shortcomings of narrowly curated datasets have also become the highlight of more recent research.
A notable effort in this respect is the Abstract Reasoning Corpus developed by Keras founder Francois Chollet.
What metrics can we use?
Could machines be said to understand differently from humans?
Mitchell writes in her paper.
What made this specific study interesting was the broad range of perspectives brought together to tackle this complicated topic.
When I first got into AI, there was a real interdisciplinary feel to it.
AI people attended cognitive science conferences, and vice versa.
Then statistics took over AI, and the field got less diverse, Mitchell said.
For me the perspectives from people outside AI (in psychology, neuroscience, philosophy, etc.)
And also that we really dont understand natural intelligence very well at all.
it’s possible for you to read the original articlehere.
