Perception: Consuming the World of AI

By Alexandre Winter / November 19, 2019

AI is vastly about perception. Artificial perception technologies are making significant advancements in the world. We believe that artificial perception is at a turning point where new sensors, approaches and data sources will soon create a Cambrian explosion of products and applications. We are creating the Perceptive AI Venture Studio to help generate these, now.

Artificial Intelligence is the art of building machines that reproduce human intelligence. Human intelligence is made of several parts – cognition, reasoning, and perception. Perception is the job of the parietal and occipital lobes of the brain. It is commonly estimated that 30% of our brain is used exclusively for vision, 8% for touch, 8% for taste and only 3% for hearing. Roughly 60% of our brain is dedicated to perception.

We can safely say that whoever solves the problem of artificial perception will win the race to achieve true Artificial General Intelligence (AGI.) 

The scientific community has been focused on questions connected to perception for the past 3-5 years. In 2018, 60% of AI research papers were related to computer vision. Computers now outperform humans in vision tasks such as face recognition and radiology.

Many tasks are still not solved in application. While significant advancements have been made in video analysis, it is still inferior to human perception in certain fields. For example, self-driving cars still require a variety of expensive sensors like LIDARs and radars to augment their subpar camera perception systems. The inability of drones to operate autonomously indoors or in tight spaces and robotic systems freezing in high-traffic environments are additional examples of sensing shortcomings. These systems are still not perceiving the world in which we hope they will live.

Yet sensors are becoming more powerful, accessible and affordable. Examples include neuromorphic event-based cameras that offer human-like robotic vision perspectives for drones or autonomous cars, hyperspectral imaging that extends the realm of visual perception — sensing temperature or humidity. This means, seeing through walls or clouds using radar imaging is now possible and affordable. Other impressive examples include mmWave radars that sense physiological signs by penetrating the body and sensing speed and movement, and haptic sensors with AI models that potentially reproduce touch in an ultra-realistic manner. These once aspirational feats are realities today. 

The next ten years will see a seismic growth of sensor, perception, and vision-related applications. The next wave of scalable businesses will leverage these advancements and solve the above challenges at scale moving us towards Artificial General “Perception”.

Now is the best time to start a business in this field. We are launching the Perceptive AI Venture Studio at New Lab to make this technology uprising happen. We are assembling a team of tech entrepreneurs with experience in and passion for artificial perception to create the companies that will make these cutting edge sensing features and models a reality. These companies will launch products that leverage and deploy new perception systems. We will help mature projects, validate ideas and invest in startups to get them off the ground.

More information on Perceptive AI Studio to follow in the coming weeks. 

Alexandre is a leader in the field of computer vision,  repeat technology entrepreneur, and inventor with ten patents in the field. He is currently the CEO and Co-Founder of VC funded, stealth mode consumer health tech sensor company, Norbert Health, and previously founded two other successfully acquired computer vision companies: Placemeter, which quantifies human activity in cities, and LTU Technologies, a venture backed startup pioneering large scale image search by content. Alexandre is an Entrepreneur-in-Residence at New Lab.

Close [x]
Sign Up for Our Mailing List
Receive the latest updates from our community and be the first to hear about New lab events and programming.