AI Is Better Than Us At Monitoring Images And Video

How long until it's better than us at other things, too?


You’ve undoubtedly heard exciting reports about how artificial intelligence, or AI, is rapidly changing some of the ways we do ordinary activities. According to Eirik Thorsnes, a system manager at Norway’s Uni Research Computing, artificial intelligence excels compared to humans when it comes to looking at pictures and videos.

Meeting a Growing Need

Thorsnes and his colleagues from a facility called the Centre for Data Analysis say there will be an increasing demand for artificial intelligence used to analyze images and video feeds. But, Thorsnes cautions it’s not the kind of technology a non-specialist could just plug in and expect to perform flawlessly.

He thinks artificial intelligence-driven video and imagery recognition could be especially useful in applications such as satellite images, health care and environmental analysis. However, for artificial intelligence to work as intended, it relies on complex algorithms, robust software suites and input from humans who are familiar with both those components.

Ways AI Monitoring Is Already Being Used

Experts have already figured out various ways to use AI for monitoring tasks that would be too cumbersome for humans to handle, including for planning purposes to make improvements in roadways. AI-assisted cameras were installed at a particular intersection and taught to recognize how many types of vehicles pass by each day. Once AI learns which information is important, it seems there are no limits to what it might achieve.

Such advancements are also being used to monitor traffic in a different way in business sectors. Live answering services are actually less efficient than automated services, which is why some businesses depend on virtual systems that offer an initial point of contact.

Callers can press buttons on their phone keypad or speak certain phrases when prompted to tell the system what they need. This is an example of how AI’s worthiness expands beyond video and image-based monitoring into the audio realm.

There is also an ongoing study involving a type of AI called reinforcement learning and its role in promoting pain relief based on monitoring continual patient feedback about how they’re feeling. The AI system will listen to telephone calls between a patient and therapists about pedometer step counts to offer customized treatments to encourage pain relief.

Why Is AI Better Than Humans Alone at Monitoring Things?

There have been many news stories warning that AI will replace jobs normally filled by humans, and Elon Musk is a prominent person who captured headlines recently by speaking about that very trend. It’s true that AI will likely decrease the need for certain types of labor, but why does it work particularly well for monitoring purposes?

For starters, once AI systems get trained to know what to identify, they won’t be fatigued after being presented with media that looks nearly identical. Furthermore, because AI usually depends on algorithms to spot changes over time, such systems might pick up on nuanced differences humans wouldn’t perceive.

Some analysts caution AI won’t replace valuable human intuition when looking out for suspicious activity. However, most of the footage that comes across live video or imagery feeds doesn’t get seen. There are simply too many things for a monitoring technician to watch at once, and this is why AI could be a good complement to existing techniques.

Also, modern society is accumulating massive amounts of image and video-based data, and sorting through all of it is certainly challenging for humans to do without help. As AI technology continues to evolve, scientists will probably figure out more efficient ways to handle huge collections of data with less manpower.

Even in the relatively early stages of AI’s rise, it’s evident there are some things technology can do better than we can, at least once it’s well trained. It’ll certainly be interesting to keep tabs on further developments within the capacities explained above and others. 

Looking small

Read next:

Expert Insight: 'An Effective Visualization Results From A Great Deal Of Curiosity And Exploration'