Still, there are companies developing applications that blend the best of machines and human effort in novel ways
All around us we’re seeing the rise of intelligent machines and services powered by them. New techniques in machine learning have allowed for dramatic improvements in the performance of tasks and in the completion of entirely new ones. Error rates collapsed across the board, recognition of images and speech shot up, and use cases that were once impossible became routine.
Voice interfaces, made possible through incredible advancements in speech recognition and natural language processing, have creeped into our daily lives. Siri summons private cars at our command, search queries are an “OK Google” away, Alexa sends us Amazon orders without having to browse on a screen, and cars are increasingly taking over their own wheels.
All of this amounts to relatively successful implementations of what’s often referred to as narrow artificial intelligence. Within a bounded domain of a given problem, you can leverage many of the gains in ML and AI to offer a flexible variety of potential solutions in a way that wasn’t possible before with traditional computing.
Tesla may be rolling out gradual upgrades to its autopilot feature for highway driving, and Uber may have assisted driverless cars on the roads of Pittsburgh, but I suspect it will be some time before a Model S is capable of parallel parking on a cobblestoned Lisbon hill on a raucous Friday night.
Even Siri, as any iPhone owner will know, seemingly comes with the promise of an all-powerful digital assistant capable of organizing your life and answering any question. Outside of very specific tasks working very well, most things you ask of it are met with amusing built-in replies (which are really just well-written error messages).
And when it comes to managing the entire communication flow between stakeholders and customers, you might not want a bot like Tay, the infamous self-learning AI chatbot on Twitter, taking care of customer service.
Most things are still quite beyond the reach of full intelligent automation, so it’s important to manage expectations of the scope. But there are a number of companies developing applications that blend the best of machines and human effort in novel ways.
One field that’s shown success is fraud detection, allowing businesses to scour through enormous amounts of data quickly, preventing losses and providing a better customer experience. Fraud detection companies like Feedzai have discovered that machine learning alone isn’t enough to do the job alone. By pairing it with human insight and intuition, Feedzai is able to leverage all the advantages of machine learning without compromising on accuracy.
Traffic and navigation app Waze has similarly bet on an optimal combination of human and artificial intelligence. Data is continually collected from its network of drivers, who contribute both actively (inputting their own data like traffic accidents they spot on the roads) and passively (GPS data showing real-time traffic flows). The more users contribute, the smarter the algorithms, suggestions and user experience become overall.
Mighty AI, previously Spare5, is also making artificial intelligence more intelligent by implementing human labor. The creators of the “training data as a service” (TDaas) platform are aware that while AI systems are becoming highly intelligent, they are unable to teach themselves the way humans are. That’s why Mighty AI offers companies human input from individuals who carry out “microtasks” to work with any AI system they wish to build.
In 2017, businesses are finding that artificial intelligence is creating enormous value, but still very much needs a guiding human hand to be fit for purpose. This isn’t necessarily a sign that we’re just a temporary crutch for our eventual robot overlords, either. In fact, I believe that moving forward, it won’t just be metaphorical human hands helping machines, but literally our minds which unlock the true value of technology.
In fact, the final frontier of technology will involve fully integrating the ultimate interface: our brains. Many recent successful attempts at low-bandwidth neural connections have been made, most effectively in monitoring brain activity for research. However, the race is on to reach high-bandwidth brain-to-machine connections by boosting the number of linked neurons, which will revolutionize the field of AI.
While some are skeptical of how far we can take this technology due to the brain’s high complexity, current advancements prove the contrary. DARPA, for example, the U.S. military’s scientific research and development organization, has already taken huge steps in developing technology that will allow humans to control artificial limbs using their thoughts, just like they would their own arms and legs.
As scientists and futurists continue to develop these connections, humans will be adding an artificial intelligence layer upon themselves and in turn, allowing our minds to become the AI technology. While it will be some time before we are able to write a text message with our minds, the rate of current developments indicates that integrating neural connections with machines will have far-reaching implications for the future, not only in the field of technology, but also in health, education, psychology, and much more.
The views expressed are of the author.
Geektime invites global tech and startup professionals to share their opinions and expertise with our readers. If you would like to share your point of view, please contact us at [email protected]