From the early 20th century, computer scientists have found great inspiration in the neural structures that neuroscientists were investigating.
According to Matthew Cobb in The Idea of the Brain, this is where the computational experts got the idea of thinking intelligence in a binary way. As neurons evolving and communicating together through nerve impulses, they imagined and conceived artificial systems that grow through binary feedbacks.
But current neuroscience has also shown how this metaphor can be misleading and doesn’t account fully for how human intelligence works. The brain is not a digital but analogic system, based on continuous neural signals…
Computers are seizing human cognitive abilities, but they won’t replace all human jobs.
Reaching a common agreement, taking care of other people, or triggering emotions are all very human ways to communicate and interact. In this sense, it’s not that these tasks have little chance of being replaced, it is that we need them precisely for their human presence. As, for many more years to come, a robot won’t ever have the same reassuring, human-like presence, some jobs will definitely stand the test of time.
Here are the purely human jobs that will most probably thrive in future robot-dominated societies…
We find many ways to oppose AI and human intelligence. More powerful and cheaper, machine automation will replace workers. Conversely, workers with human and flexible skills put themselves against automation.
But few dare to consider humans and AI as allies who collaborate to make better decisions and increase their impact. According to Agrawal, Goldfarb and Gans in Prediction Machines, this is a pity, because these two types of intelligence are complementary, and would allow the creation of predictive devices with unprecedented accuracy.
Human cognitive flexibility and machines predictive’ scalability would help together to make more accurate and impactful decisions. …
It is easy to assume that the most successful nations in history were also the most inventive ones (China with gunpowder, Egypt and the windmill, among others).
Yet, the central place that Europe took in the 15th century seems less due to its inventors and more due to a unique social system based on cultural and cumulative transmission. By being told to leave their family nucleus, to learn from strangers, and to rely on the knowledge of others, Europeans were forced to rely on collective intelligence.
According to Joseph Heinrich in The Weirdest People in the World, this has resulted…
It’s easy to assume that virtual communication will literally distance us from creating or maintaining human connections.
Yet, as VR technologies are growing, they are bringing a deeper social and intimate experience. By their immersive and interactive nature, they paradoxically create a real sense of presence, whether with a fictive protagonist or not.
So much so that, according to Peter Rubin in Future Presence, they question the very notion of human presence and connection. If we can relate with a virtual avatar in a digital environment, does that mean that we don’t need physical presence to experience genuine emotions? …
Customer mass targeting tools have only emerged in recent decades with the accumulation of data. Yet, in earlier times, Jill Lepore in If-Then tells the little-known story of a company that had pioneered the interpretation of data to understand and influence minds.
Without taking a moral look at it, this political legacy reveals how ambiguous all the marketing techniques was then seen by the public: either as legitimate ways to change their minds, or as malicious attempts to manipulate their opinions.
Now, new technologies give marketers the power more than ever to put consumers’ mind in their favor. And yet…
If deep learning neural networks have shown progress in word recognition through Word2vec models, they still reveal flaws in their learning system. In particular, algorithms fed by mathematical correlations fail to understand semantic associations.
According to Hofstadter and Sander in Surfaces and Essences, these machines lack a human sense of analogy. In comparison, human speakers can apply words or expressions in very different contexts, and extract very different meanings from them. Words don’t belong to rigid categories but easily change and evolve.
This great plasticity of meaning fuels language with many stories. These narratives, otherwise known as “common sense” (e.g…
In their quest for cognitively powered machines, researchers have been largely inspired by the structure of the human brain. By creating neural networks with successive layers, they have made them process and interpret exponentially large amounts of data. However, it seems they are missing a part of the story.
According to Joseph Heinrich in The Secret of Our Success, the human brain has not stood out by individual intelligence but by knowledge and know-how cumulating through the ages. …
With their great deterministic and scalable structure, neural network architectures can accumulate a lot of data in an exponential rate.
Yet, what makes them learning so fast is probably what prevents them from a deeper understanding of their environment.
In fact, every part of their structure is already determined by their function, and thus leaves no room for organic complexity. In contrast, each description of the human brain cannot predict the resulting behavior.
Michael S. Gazzaniga, in “the Consciousness Instinct”, brings this back to the semantic and arbitrary part of the DNA: every time RNA translates DNA there is always…
Through academic contests these past 20 years(Imagenet, Alphago challenge, Word2Vec…), AI has shown huge successes in specific fields.
Whether object, voice or word recognition, deep learning algorithms from major tech companies have shown accuracy close to perfection.
But these results hide imperfections that could soon hinder their progress. According to Melanie Mitchell in A Guide for thinking humans, AI lacks the basis of a generalist intelligence that understands and adapts to situations, and that could significantly slow down data learning.
Here are four cognitive skills that machines lack, and the skills they need to pass these obstacles.