Artificial Intelligence Applications and Examples
Artificial Intelligence: Applications
We hear about impressive achievements related to Machine Learning and Artificial Intelligence every other day now. Even so, most of us are still dubious about the applicability of AI in our space, and with good reason. Take the remarkable performance of an AI-based system on a database with millions of images for instance. This won't always translate into the same rate of success “in the wild” where context, angles, lighting conditions and image resolution may be very different. When somebody performs a task well, we naturally assume that the person has some competence in related tasks. But AI systems are trained to do specific jobs, and their knowledge typically does not generalize. The most significant source of exaggerated claims and confusion regarding AI’s progress stems from a false belief that a system’s narrow understanding implies broader understanding. We are far from machines that exhibit general intelligence across diverse domains.
Machine learning systems too can hardly ever replace the entire job, process, or business model. They are devised to complement our work. Making their work ever more valuable. The most effective rule for the new division of labor is not “give all tasks to the machine.” Instead, if a process requires ten steps, one or two of them may become automated while the rest become more valuable for humans to do.
The applications today are endless. We are already aware of AI-based systems categorizing images and videos in many cloud services. The next step is implementing the same in smart camera feeds . Speech-based assistants are another common AI win. They are increasingly becoming a part of our lives, as they play music and control primary devices in our smart homes. Dialogue is a fundamental human tool we often take for granted. Intelligent devices you can have a conversation with are a revolution underway. Speech-based assistants are getting better every day at serving us. Siri, Cortana, and Alexa are always with you and always on. This is again another victory of the smartphone. But we also want this feature in our vehicles. We need less and less cloud and instead, local processing of voice. Lesser bandwidth costs and more privacy. Again hardware will provide us with all this in a year or two.
For now, we have “chatbots” and automated virtual assistants answering customer service calls. In fact, for pizza delivery, self-driving cars are being tested. AI devices, such as drones, are being adapted for all kinds of creative uses. Researchers in Australia have developed flying drones capable of identifying sharks near surfers and swimmers, sending out electrical impulses that deter and irritate sharks from entering populated areas and amplifying warnings to beachgoers through an onboard loudspeaker. AI can point out the disease with up to ninety-eight percent accuracy. Via smartphone, farmers are using this technology to take pictures of diseased crops. Based on severity, phase timing and plan specifications in the construction industry, AI is being used to assist project managers to track the most egregious potential malfunctions. With a laser-like focus on quality and safety, this helps keep projects on budget and on time. The retail industry too has understood the impact it can wield. Burberry, a British fashion icon requested and uploaded plenty of data on their customer's buying habits. This allows frontline retail clerks to make immediate recommendations, based on what customers purchased in the past to complement client selections. The Intelligent system has given birth to a personalized shopping experience that has proven enormously successful. These quantum jumps in technological advances present both challenges and opportunities.
For organizations looking to put ML to use today, there are three pieces of good news. Firstly, ML skills are increasing fast. The world still doesn't have nearly enough machine learning experts and data scientists, but the demand is being addressed by universities as well as educational resources, online. Fast.ai, Coursera and Udacity do much more than teaching introductory concepts. They can get motivated, smart students to the point of being able to create industrial-grade ML deployments. Secondly, another welcome development is that the necessary hardware and algorithms for modern AI can be rented or bought as needed. Amazon, Google, Salesforce, Microsoft, and others are making powerful AI infrastructure available through the cloud. The competition in the market means that companies that want to deploy or experiment with ML will see a lot more capabilities available at far lower prices over time. Finally, the most underappreciated piece of good news is that you might not require as much data as you think to begin making good use of AI. It seems logical to conclude that the organization with the most data will win, as machine learning systems improve their performance proportional to the amount of data they're given. By “win,” if you’re looking to “dominate the space for a certain application like speech recognition or ad targeting,” then yes. But if you define success as markedly improving performance, then enough data will prove easy enough to obtain. Whether you know it or not, you are ready for AI-based systems. But in the end, how successfully you helm disruptive technology will boil down to your leadership.