Instead of giving credence to fear-mongering books and films about artificially intelligent beings taking over the world and turning on humans (The Terminator, anyone?), it’s time we took a hard look at the proven benefits and potential risks of some real-world applications of artificial intelligence.
People typically associate artificial intelligence with the technology sector; however, AI is increasingly being used throughout industries that aren’t strictly tech-based.
Benefits of artificial intelligence
Throughout the last half century, artificial intelligence has become an important part for success across various sectors. For instance, it has become commonplace to see AI used in:
These industries (as well as others) have opened the door to increased opportunities and advancements that AI can bring as a complement or supplement to human thought and labor.
The healthcare field has many subsets that have been positively impacted by artificial intelligence. In particular, AI has influenced radiology and digital consultation.
AI in radiology
Radiology is a specialty that uses diagnostic (medical) imaging to diagnose and treat diseases in patients. The purpose of diagnostic imaging is to see inside a patient’s body, determine if there is an issue at hand, and diagnose treatment or further testing. Diagnostic imaging includes x-rays, CT scans, PET scans, MRIs, and ultrasounds.
Image courtesy of Medicalview.org
AI factors into radiology by using deep learning algorithms – specifically convolutional neural networks (CNN) – to help with detecting and diagnosing diseases. CNN is most commonly associated with analyzing visual imagery, hence its relation to and assistance with diagnostic imaging.
For now, AI’s purpose within radiology is to act as a second set of eyes – or a second opinion – that can aid with the level of accuracy of an initial diagnosis made by a physician. Additionally, AI can assist with reducing the time and human error of missing slight changes in diagnostic images – time that, if wasted, could be detrimental to patient wellness.
AI in digital consultation
Digital consultation tools (or “online doctors”) record illness symptoms that users input online. The artificially intelligent program then draws assumptions of diagnoses and solutions based on other user inputs and experiences.
With most clinics and doctors’ offices closing before 6 p.m., working individuals find it difficult to carve time out during the workweek to visit their physician. Similarly, most people would prefer not having to take off of work to go to the doctor. Not only is it a hassle, but it interferes with day-to-day processes like a full-time job.
To remedy this, people have begun turning to digital consultation tools in order to get help with their health concerns without ever stepping foot into a doctor’s office. Though websites like WebMD Symptom Checker have existed for some time, it does not use artificial intelligence to draw conclusions about health issues the same way digital consultation AIs do. However, more advanced online symptom-checking websites have sprung up with the increased need and desire for remote medical assistance.
One impressive digital consultation tool that utilizes AI is Buoy, an interactive symptom-checking chatbot that prompts the user with more questions about symptoms based on the initial user selection. For increased accuracy, Buoy begins by asking the user their gender, age, and most bothersome symptom. With these things in mind, Buoy continues prompting the user based on their initial input – thus attempting to develop a knowledge base similar to a human physician. The same way a doctor may ask her patient to indicate related areas of pain and concern, Buoy asks this of the patient as well.
For instance, if the user indicates they have a headache, Buoy will ask questions about head pain and related areas: face, eyes, neck, throat, and so on. To mimic human conversation, Buoy responds with phrases like: “Alright, let’s figure this out together.” and similar dialogue-esque phrasing that seems to come from a human being, not an AI chatbot.
At the end of the user’s “consultation,” Buoy thinks about the user’s inputs and connects the dots, thus leading them to seek additional help (if necessary) based on its diagnosis. Not only does Buoy provide the user with an illness that most closely matches the patient’s symptoms, it also provides suggestions on how soon to see a primary care physician and nearby clinic options based on the user’s zip code or computer location.
What takes Buoy a step further from symptom-checkers of the past is the ability to “converse” back-and-forth with an AI consultant and get results in a matter of minutes. At the very end of the consultation, Buoy asks the patient what they think is most likely the right answer, thus trusting human judgment in conjunction with the judgment call the AI chatbot has already made.
AI has been implemented in manufacturing for at least the past 50 years. As early as the 1960s, General Motors added Unimate – the first industrial robot – to an assembly line at a plant in New Jersey. Its tasks included welding and transporting die castings on to cars, tasks deemed unsafe tasks for humans. In vehicle manufacturing plants, industrial robots are a necessity for producing large quantities of materials.
Presently, countless large and small manufacturing companies welcome industrialized robots programmed with AI to assist with laborious, unsafe, or repetitive tasks. In some cases, machines have replaced human laborers because they can be programmed to do a task at near-perfect accuracy with lower levels of burnout or error than their human counterparts. In other cases, machines simply assist humans with increasing output at a higher standard.
On top of improved accuracy and speed, industrialized machines can be programmed to identify recurring patterns that may result in issues with quality of output – patterns a human may not catch for quite some time. Additionally, some machines have been designed with image recognition (computer vision) which gives them the ability to see, process, and inspect materials for flaws more successfully than the naked eye.
Overall, adding artificial intelligence to manufacturing has benefitted companies in the areas of accuracy, yield, cost, and quality control.
Image courtesy of Acieta.com
Banking and finance has dabbled in automated systems following the invention of ATMs in the late 1960s. First appearing in Britain, the United States quickly caught on to the “money machine” trend in 1969, something people feared would eliminate the job of a bank teller. Presently, AI, machine learning, and big data are used in the financial sector in ways that benefit both customers and the financial institutions themselves.
In 2016, Bank of America debuted a virtual assistant named Erica. Dubbed a “virtual financial assistant,” Erica can take care of customer needs and concerns in a few key areas such as tracking users’ spending habits, managing bill payments, and locking debit cards, to name a few.
Image courtesy of BankofAmerica.com
Another example is JPMorgan Chase who has made significant investments in technology, AI, and machine learning. In 2019, Chase introduced the JPM Coin, a “digital coin [that represents] a fiat currency” using blockchain technology for instant “transfer of payments between institutional accounts.” Chase is the first bank to create its own cryptocurrency. The pilot Coin program will only be available for JMP’s institutional clients for the time being.
JPMorgan Chase also has tech tools designed to analyze legal documents while extracting data – a task that is usually done by manual review. However, implementing technology that can extract data just by viewing and “comprehending” text saves on time and manpower as such a task normally takes over 350,000 hours to complete.
Customer service automation has been progressing significantly over the past decade. With the introduction of chatbots to assist in decision-making on websites such as Facebook, people can message and “interact” with chatbot assistants in real time. Facebook chatbots are powered by artificial intelligence – or, depending on the case – pre-programmed responses that the system understands based off of a human-directed inquiry.
Chatbots function similarly to customer service representatives (CSR) for base-level questions by using natural language processing (NLP) to respond to customer concerns in a timely and accurate manner. NLP is a subset of artificial intelligence that helps computers interpret and respond to human language.
Automating customer service can help companies remain accessible during off-hours. For instance, if a company’s working hours are 8 a.m.-6 p.m. (CT), but someone from a different part of the country or different country altogether needs to contact a CSR, there may never be a viable time that works for both the customer and the company; however, with the accessibility that is provided by chatbots as CSRs, responding to questions and concerns of high importance can be done with a customer’s schedule in mind.
AI in e-commerce
To keep up with online-only retailers like Amazon, many brick and mortar stores have implemented artificial intelligence chatbots as a way to communicate with customers or assist in customer buying decisions. Oftentimes, these chatbots live on Facebook and can be accessed by the customer prompting the chatbot “awake.” This benefits Facebook because its Messenger service is the means by which retailers encourage their customers to reach out.
What makes each chatbot unique is the fact that some of them are completely automated; some are run by a human with a window of time when the CSR will respond; and some reply instantly but then direct the customer to reach out via phone call or email.
Some retailers, such as Sephora, not only prompt the customer with a pre-written greeting message, but also allow the customer to make self-driven decisions about using their “try on looks” tool or reaching out to a human customer service agent.
Because customers are inclined to prefer different levels of human CSR interaction versus AI CSR interaction, companies may find more or less success with their approach depending on who they ask.
Potential risks of artificial intelligence
Naturally, with all technological advancements come setbacks. While still abstaining from becoming modern-day Luddites, it’s important to discuss the potentially harmful implications that AI integration can and does have on the aforementioned industries.
Invasion of privacy
Naturally, there have been fears about artificial intelligence “spying” on its users – and for good reason. With home voice assistants like Google Home and Amazon Alexa listening in to user conversations, people’s fears about being tracked based on what they say, think, or search online are understandably valid.
To put things in perspective, over the past few years, China has implemented a social credit system that heavily tracks and evaluates people’s social credit based on a combination of mostly minor offenses such as jaywalking, smoking in non-smoking areas, not putting one’s dog on a leash, running stop signs, not paying taxes on time, or not giving up one’s seat on a train to someone else. This social credit system tracks citizens by their internet activities, including social media profiles, financial records, private messages, health background, dating history, and consumption of media (books, TV, video games.)
Image courtesy of Kevin Hong
Obviously this system has negative implications for societal “wrongdoers.” People with low social credit have been barred from purchasing plane tickets to travel, enrolling their children in elite schools (despite their children’s skills and abilities), or even leaving the country in some circumstances. Conversely, those with high social credit scores benefit handsomely from the system. For instance, rule-abiders are rewarded with easier access to loans, discounts on utility bills, increased internet speed, and increased access to coveted property.
If a social credit system is applied globally, the same adverse effects happening to Chinese citizens with poor social credit will occur elsewhere. A system of this nature is not only used to harvest personal information from the country’s citizens, but it can be used for purposes of discrimination based on race, religion, and class.
Loss of jobs
In some sectors, job loss is imminent. With external factors taken into consideration, some industries may find that AI-programmed employees are successful and more productive at higher rates than their human counterparts. When dealing with machines, a few human-only concerns can be eliminated: personal and emotional situations; lethargy and exhaustion; and boredom and distraction.
It is evident that AI-fueled machines do not have the emotional capacity that humans have despite many advancements the field (including the creation of realistic humanoid bots with facial expressions and the ability to converse.) Because of this, mechanized employees are ideal given that they cannot possibly be affected by emotional occurrences inside or out of the workplace.
Additionally, machines are often powered by electricity (via a plug and socket outlet) or battery, either rechargeable or disposable. Thus, machines programmed with AI and powered by electricity or batteries will not face “brain drain” like a human might in the afternoon of a long day at work. AI-fueled machines do not get tired and will not slow down simply because they’re losing “steam” as the work day winds down.
In conjunction, AI-programmed machines won’t find repetitive tasks such as packing, sealing, and stamping boxes for shipment to be boring or tedious. All the machines know is what they are programmed to do and thus will not be distracted by outside interests like joking with coworkers, taking personal calls or responding to texts, or smoking and coffee breaks.
GIF courtesy of Amazon via Medium.com
Though machine error can occur (as a result of human error in terms of programming), human error in a workplace can lead to greater losses in revenue for a company over time. In essence, AI automated employees are preferred by many manufacturers given their programmability and product output over a period of time.
Mechanizing for purposes of war
Because AI still has to be programmed by humans, there is a highly realistic concern that some people desire using AI to program machines to cause harm to humans or act in a weaponized fashion. Many world leaders believe that AI will become more ingrained in society in the coming years, thus providing opportunities to integrate artificial intelligence into military force by means of autonomous weaponry.
It is logical to assume that some countries or individuals may want to utilize artificial intelligence to increase the likelihood of winning a hypothetical war. If machines are programmed with intent to kill, they become a danger for rival countries who may not have the same technological advancements.
Image courtesy of Wired.com
In 2017, Russian president Vladimir Putin stated: “Artificial intelligence is the future, not only for Russia, but for all humankind. It comes with enormous opportunities, but also threats that are difficult to predict. Whoever becomes the leader in this sphere will become the ruler of the world.” This statement highlights one world leader’s viewpoint on where artificial intelligence may be headed in the near future.
A final concern regarding AI-programmed machines for military usage is what might happen if weaponized machines gain “a mind of their own.” If a machine once programmed to kill and be used solely for purposes of war goes rogue, it may not be able to discern “enemy” from “friend.” Granted, this idea is theoretical and more far-fetched than tangible; however, with the rapid advancements in the field of AI, it could happen someday.
What should we expect in the long run?
Implementing artificial intelligence as a supplement to (rather than a replacement for) human intelligence seems to be the trend in AI. It’s true – AI can and likely will eliminate some easily automated jobs in the near future. But instead of worrying about robots taking over, it is evident to see that artificial intelligence boasts more benefits than risks.
Regardless of how many industries decide to integrate AI as an addition to human skill, thought, and labor, the common denominator is clear: human beings remain the springboard off of which AI has existed, does exist, and will continue to exist in the future.
Still not convinced of the benefits artificial intelligence can bring to the world? Check out the history of artificial intelligence to become more acquainted with AI discoveries and how they've helped shape life as we know it.