BACKGROUND IMAGE: agsandrew/stock.adobe.com
Companies are competing in a billion dollar arms race to implement artificial intelligence, but experts are beginning to see cracks in the relentless pursuit. This is primarily because, so far, big AI promises have produced little concrete innovation. Most AI technology still struggles to complete tasks that a four-year-old child accomplishes effortlessly.
The current "intelligent" systems are shortcuts that detect patterns based on statistical methods but still have no understanding of what they have detected. This understanding and the ability to mimic human task completion -- what developers refer to as artificial general intelligence -- is a long way from full development.
Rather than focus on a general goal to improve artificial intelligence, CIOs and businesses should start paying attention to -- and investing in -- augmented intelligence technology. Augmented intelligence technology is a form of artificial intelligence that does not seek to replace humans, but instead seeks to assist humans with their work. This makes augmented intelligence a concrete and ROI-accessible alternative to AI.
No replacement for humans
The automotive industry is a great example of the potential of AI, but also serves as a great example of AI's flaws. Mercedes, Toyota and Nissan were among the companies that touted their self-driving car concepts, but the biggest question to come from the 2018 Consumer Electronics Show in Las Vegas was, "Where are the autonomous vehicles?"
Tesla's Autopilot is not a full replacement for self-driving, and Uber's vision was crushed after a fatal accident involving one of its self-driving cars. Tesla's Autopilot, on the other hand, recently crashed into a stopped firetruck because the AI is trained to ignore them. Trying to lean toward the side of safety, Waymo's cars try to abide every single traffic rule but create inferior systems that would fail basic driving tasks.
Augmented intelligence overrides the critical problem of AI adoption: that the systems are reactive, not proactive. A human driver has the ability to make educated guesses about what will happen on the road 10 seconds in the future, whereas the current AI systems have no way to make such predictions. It's here that augmented intelligence could make a difference. The trick is to train the system so that it accurately recognizes the subjects it is trained on, as well as offer reasonable performance when it comes to unforeseen road hazards.
An extra 'Eye'
Discoperi Inc.'s System 'Eye' is reflective of augmented technology as a first step to complete AI adoption. The Eye is a camera placed on a vehicle that combines the data from several cars that allows the system to form a grid to assist all nearby drivers. This augmented intelligence doesn't operate the vehicle, but instead assists the primary, human operator by detecting objects and behavior patterns to recognize when a driver is breaking from normal patterns and driving dangerously -- hoping to reduce fatal accidents.
The Eye demonstrated great accuracy when detecting objects on the road, but the AI's primary task is to build behavior patterns. Of course, there are several parameters involved, such as where an event happens, under what conditions and whether there are pedestrians on the road. The system checks what's normal under these circumstances to what is currently happening, and if it is beyond a certain threshold, it will send an alert to all cars within that proximity.
While access to information about every car on the road sounds like a privacy nightmare, Discoperi has already taken steps to ensure privacy as well as give users full control over their data by storing the data on a blockchain.
Privacy might seem more like a problem for augmented intelligence because it involves human input, while artificial intelligence is theoretically fully autonomous. But due to the real shortcomings in AI, many companies have already used humans behind the scenes to complete AI's job where it failed, raising privacy concerns in AI.
Augmented intelligence capabilities will benefit industries -- perhaps even more than traditional AI would. In healthcare, an AI algorithm can analyze a patient's symptoms and vital signs, compare it with the history of the patient, their family and millions of other patients, and provide possible diagnoses a for a doctor to analyze. In education, AI can track the progress of the students and help teachers understand what topics and which students need more attention.
Augmented intelligence can contribute to streamlining business processes and aid human workers in a concrete, helpful and cost-effective manner.
AI has undergone two winters by now, and the next one is looming. But this time, AI has made far too much progress to be forgotten again. CIOs can't continue investing in stalled technology, and they should also resist giving up the benefits of AI technology. We need to find the correct balance and, for now, it seems the best way would be to combine AI with human intelligence and form augmented intelligence technology.