Home > Blogs > From Customer Service to Radiology: Applications of AI
Andrew Ng, Co-founder of Coursera, Adjunct Professor of Stanford University, and formerly head of Baidu AI Group/Google Brain, recently received an interesting query from a radiologist. The radiologist asked Andrew Ng if he should switch professions because of the increasing possibility that artificial intelligence might replace radiologists in a few years.
His query may seem far-fetched to some, but in the backdrop of machines reaching levels of intelligence wherein they outperform humans at most tasks, it is not improbable.
The first wave of jobs to be automated in the industrialization era were those which involved mechanical actions. Tasks requiring repeated motions of human arms and limbs were automated using mechanical assembly of nuts, bolts and minimal computation. As the computing prowess of machines increased, the ability of machines to accomplish complex tasks increased. However, what had remained elusive (apart from the limited confines of academia) were machines performing tasks that required application of intelligence and not just following a set of instructions.
Let us take the simple example of the task of translation. A crude approach devoid of learning and intelligence, would result in the software translating the sentences “I object”, and “A red object” by using the same meaning of the word “object”, resulting in incorrect translation. Words mean differently based on their usage, and context. A proper translation machine will be required to learn the different type of usages of the words, and since translation is done for the humans, to return the translation in a form fit for human consumption. That is, a translation of English sentence to a Hindi sentence should come out a cohesive sentence with proper grammar, not as a jumble of words. These tasks come very naturally to humans but a machine can emulate such behaviour to a large extent with the help of complex algorithms and huge amounts of data.
With the advancements in the fields of machine learning, computer hardware (more specifically, microprocessors), and cloud computing in the past two decades, machines can now effectively take part in a feedback loop by crunching huge amount of data in a relatively short amount time, learn from that data and get better iteratively. Taking forward our translation example, these advancements have made google translate possible. This has also allowed accomplishment of several more such complex tasks. Several enterprises have found interesting applications of artificial intelligence in different fields. Here are some examples:
1. ROSS Intelligence: Application of AI in Legal Services
The usual approach of finding answers on the Internet is the keyword based search. While we may have adjusted to this approach it does not reflect the real world scenario that is more semantic. People exchange information in a meeting not by randomly shouting a word and wait for their colleagues to give appropriate response, but by asking relevant questions. And nowhere is this need felt more acutely than the legal services.
While you and I may ask Siri “what is the capital of Sri Lanka”, enterprises’ expectations are something along the lines of “Draft a contract pursuant to the agreed upon conditions, meeting the compliance standards and ensure it is legally tenable in a court of law”. For such tasks, help is at hand from the ROSS Intelligence System.
With the help of IBM Watson in the backend, a lawyer can ask legal questions to the system in the form of normal English sentences rather than obtuse database queries. This system utilizes natural language processing techniques to check for English sentences in the legal documents which are relevant to legal query at hand. The system will parse the query sentence, understand the context, sentiment and data points, and return relevant legal documents to the lawyer. Prior judgements, precedents, amendments, notifications, and data filtered from millions of documents containing information covering several decades will be served to the lawyer within minutes.
ROSS Intelligence System can take the drudgery out of the legal services and help lawyers serve their clients more effectively. As per ROSS Intelligence’s website, the system can be set up in minutes, and since queries are direct questions, using the system doesn’t require extensive training. ROSS Intelligence has already entered into partnership with world’s top law firms like Latham & Watkins, and leading universities like Vanderbilt Law School.
2. Cogito: Artificial Intelligence to the Aid of Tele-callers
A customer calls his insurance company to enquire about an issue related to his premiums. The agent on call tries his best to address the customer’s queries but due to communication gaps, both the customer and the agent return from the call dissatisfied and irritated. This next product can end such annoying experience frequently related with service related phone calls. Cogito’s Product Dialog provides real time behavioral analytics of a customer to telecommunication agents while they are on calls. It provides insights into the customer’s “feelings” (such as irritation, restlessness, or openness) during the service call based on auditory cues and speaking behavior. Such information is then displayed to agents, along with suggested remedial measures which the agents can use amend their conversation style and reform their pitch. For example, if the agent speaks too fast, the system will provide feedback that the customer is feeling confused. The system also provides the ability to the agents’ supervisor to monitor the calls in real time and takeover from the agent if they are struggling.
Several major players in the field of insurance, health services, and financial services such as Humana, CareFirst and Zurich Insurance have already partnered with Cogito. As per a report by Radiant Insights, customer service industry is going to hit 84.7 billion dollars by 2020. A product like Cogito’s Dialog can ensure that those dollars are deployed to maximum impact. It can help customer services get better in customer retention and satisfaction, as well as attracting new customers. Maybe you won’t feel like disconnecting the call of a telemarketer mid-way next time, and you can thank Cogito for that. A more emotionally intelligent agent, will be a better agent.
3. Zebra Medical Vision: Application of AI in Radiology
Perhaps the radiologist mentioned earlier had this company in mind when he wrote to Andrew Ng. Zebra Medical Vision, an Israel-based company, uses artificial intelligence techniques to predict diseases based on medical images. The company uses Deep Learning, a branch of machine learning derived from neural network, to analyze radiological images and check for early signs, or onset of cardiovascular events, liver diseases, or lung infections. The efficacy of artificial intelligence is dependent on the data which it is trained on, and Zebra Medical vision boasts of one of the largest anonymized repository of medical images. These images have helped them train their deep learning algorithm to a degree where it can even outperform the humans.
The service can also be used by the insurance companies to check the risk level of the person they are insuring. They tested their algorithms on historical data, and the results were very promising. Zebra Medical Vision has already attracted prominent partners in the field of healthcare like Intermountain Healthcare, Telerad Tech, Mahajan Imaging, University Of Virginia Health Systems, and more.
4. Crystal: Improving business communication
How do you write a mail to someone with the best chance of getting a response, or having the desired impact? Start-up Crystal helps with that. This site helps people draft better emails by analyzing the publicly available data of the person you are writing to. Based on the person’s social media posts, his past interactions, Crystal discovers personality traits, and the most suitable strategy for communication with the intended receiver. The strategy is delivered as simple recommendations to the author of the mail. An example of such recommendation can be “Include bullet points”, “Avoid exclamation marks and emoticons”, “Be direct and avoid clichés”. This tool not only helps the sender but is also a boon to the receiver as they will receive more meaningful messages, tailored to their preferred style.
These are just four examples of the real life application of Artificial Intelligence and how it is affecting businesses and lives directly. Zebra Medical vision claims that their invention can save up to 500,000 lives annually. Insurance provider Humana saw a 28% increase in customer satisfaction, and 63% improvement in employee engagement after they partnered with Cogito. If enterprises have to remain relevant to customers, or attract people to their offerings, they have to leverage the power of artificial intelligence in their services and products. The impact will not only be visible in their quarterly statements, but also in the lives of people across the world.
About the Author
Arvind is a Post Graduate from IIM, Lucknow with about 20 years' experience in diverse roles driving growth across various businesses and geographies. His last 10 years have been in leadership roles advocating technology in Learning, Training and Assessment across a wide range of customers across the globe. He has a strong track record in establishing processes that drive revenue growth. He comes with a rounded experience having worked in large firms and in smaller set-ups as well leveraging on the learnings from both ends of the spectrum. He started his career in the FMCG and trading sectors before his twin interests in learning and technology drew him towards this field. For the past few years he has devoted himself to various aspects of educational delivery evangelizing technology solutions. He has a firm belief that technology is the catalyst that will ultimately equalize the quality of learning.