Unlocking the Power of Artificial Intelligence (AI): Navigating the Benefits and Risks

Alexa, Chatbots, Self-driving cars. What do all 3 of these things have in common? These tools use some form of artificial intelligence (AI). AI is a ubiquitous staple in our lives. Voice assistant Siri and facial recognition systems are other common examples of artificial intelligence tools. In this blog post, I will discuss artificial intelligence (AI) from a general overview and its prevalent role in various sectors. To illustrate its growing longevity, about 1/4 of U.S. businesses have already begun to adopt artificial intelligence (AI), and about 1/2 of companies have already implemented AI-enabled tools in their respective products and services.

My innate curiosity inspired me to learn and write about this topic. My unwavering desire to spread knowledge to people to empower themselves is the constant driving force that propels my "why" for writing. The second reason I chose to write about AI is to understand it and its terminology better. A basic understanding of its purpose and functions will enable me to stay abreast of important news updates and to be an informed consumer of information in the digital age. Undoubtedly, AI has been and will continue to be an integral part of our lives on an individual and societal level in the long run.

Voice assistant Alexa employs AI to understand and respond to voice commands.

To kick things off, artificial intelligence (AI) is a subfield of computer science that utilizes computational models and systems to imitate fundamental aspects of human behavior, such as learning and reasoning. We incorporate AI into our daily lives in both overt and subtle ways. For example, Netflix uses machine learning, a type of artificial intelligence (AI) that creates algorithms based on existing data that can, in turn, make more intelligent predictions and decisions without the need for computational programming rules. To elaborate on my example, Netflix employs machine learning by using viewers' past watch history as baseline data to make tailored recommendations for potential movies and TV shows that viewers might be interested in watching. 

One machine learning method is supervised learning, a category of machine learning that utilizes labeled datasets(collection of categorized data) to organize and classify data. For example, a person can train a computer to recognize dogs. Humans would have to feed images of dogs and unrelated objects into computer machines. Over time, they can ascertain important characteristics and patterns that can differentiate between dogs and other objects, and in turn, they can produce images of dogs.

Netflix uses AI to make personalized viewing recommendations for users based on their viewing history.

Another type of artificial intelligence (AI) is a neural network. It is a computational model that mirrors the human brain's structure and functioning. Some examples of neural networks are facial recognition systems built into our smartphones, which can unlock our smartphone devices, and MRI/CT imaging, which can detect malignant tumors or other abnormalities in medical imaging.

A pictorial representation of artificial intelligence

Moreover, Generative artificial intelligence (AI) is a subtype of AI that can instantaneously create images, text, and audio to create aesthetically pleasing and creative content. An example of generative AI is Grammarly. It employs AI to generate grammatically correct language and stylistically sound writing. It provides tailored feedback and suggestions to improve one's writing and communication and enhances productivity. Another example of generative AI is the musical platform Aiva, which creates innovative musical pieces conducive to a particular genre or ambiance. The CPT-3 platform can produce compelling stories and texts using a specific prompt. Aside from the fine arts and literary areas, generative AI has been used in the pharmaceutical arena to create new medicines from molecular structures and in the video gaming industry to create vast, limitless variations of video games.

In 1950, renowned British computer scientist and mathematician Alan Turing conceptualized the "Turing Test," which measures a computer machine's ability to show human-like qualities. Researchers conduct the Turing Test using a human user engaging in conversational dialogue with another human and a computer machine. A computer machine passes the Turing Test if a human judge cannot soundly differentiate between a computer machine and a human's conversational dialogue.

German-American scientist Joseph Weizenbaum developed the first natural language processing (NLP) computer program called the ELIZA in 1966. It simulated psychotherapy, most notably Rogerian therapy. The ELIZA computer program aimed to demonstrate that interaction between humans and machines is superficial. Weizenbaum noticed that his assistant formed a personal relationship with the ELIZA program. This observation subsequently prompted Weizenbaum to state that humans can grow and become attached to computer machines despite their limited scope. People tend to ascribe human-like traits to technological devices even though they are fully aware that they are limited in scope, known as the "ELIZA effect."

The hype and buzz around artificial intelligence (AI) is palpable. Various industries are implementing it. Like everything else, AI presents various potential pros and cons. From a mental health perspective, the increasing use and reliance on AI can widen people's access to therapeutic care in chatbots and digital health therapeutic applications, or "apps" for short. There has been a lack of mental health providers, and the number of individuals who struggle with mental health disorders is at an astronomical rate. In other words, the demand is higher than the supply. AI can alleviate this burden for mental health practitioners and reduce their workload from performing mundane, administrative tasks to dedicating more of their time to the human side of therapy. I am confident that AI can be a helpful supplement in helping mental health practitioners reduce their workload from performing mundane, administrative tasks, such as report writing/documentation, to ultimately focus on providing higher quality clinical care to clients/patients. Some existing methods of how AI is employed are through the use of biodata to gather info on users' medical history to make sound predictions on users' susceptibility to becoming diagnosed with mental health disorders. Another existing method of using AI in healthcare is digital health therapeutics, software programs that deliver evidence-based therapies and can treat various clinical disorders.

ChatGPT could be a game changer in the AI field.

Artificial intelligence is, unfortunately, not without its fair share of issues, and it's crucial to be mindful of this. Biases and inequity are the two biggest problems that plague artificial intelligence and also in our society as well. It can wreak havoc in individuals' lives by replacing or displacing their jobs or jeopardizing their overall well-being, namely their reputation or sense of privacy of their personal information. Even long before the advent of artificial intelligence, humans have held and continue to hold longstanding biases against disadvantaged groups, which can produce cascading effects.

There are countless other examples of how bias infiltrated AI. In the criminal justice field, the recidivism risk of black offenders is significantly higher than it is for white criminal offenders. This type of negative bias can, in turn, place them at a disadvantaged place in society in that it would cause them to endure discrimination and be racially profiled at the mercy of the policing system. Another example of bias that has infiltrated AI and taken place on social media platforms is inaccurately labeling all Black people's tweets as a form of "hate speech" on Twitter.

ChatGPT user interface

Chatbots, conversational agents that converse and interact with human users, are another staple of artificial intelligence (AI) that is heavily prevalent in digital therapeutic spaces and the customer service arena. There are some potential advantages and downsides to the chatbot feature. One advantage of the chatbot feature is that it can scale mental health support and services to reach a broader population. Another possible benefit of chatbots is that they are available to users 24/7, which makes them highly convenient. Chatbots, on the other hand, can present some downsides. A disadvantage could be little to no regulation regarding data monitoring, which can instill mistrust in people. On the flip side, chatbots can make people overreliant on their use. Too much of anything can be harmful. People may make the erroneous mistake that chatbots can fully replace human therapists. Chatbots do not possess the wide-ranging therapeutic skills that clinicians do. There are a few examples of how chatbots have inflicted harm on others. The Chatbot feature named Lawbot was initially designed by university students from Cambridge University to aid victims of sexual assault and to seek legal recourse. It dropped the ball in terms of its efficacy in terms of inflicting emotional harm, being emotionally insensitive to victims, and placing rigid rules on the criteria of what is considered a crime, which can deter victims from seeking help in the first place. Another way the Lawbot chatbot feature failed was that it prompted victims to contact the police rather than seek counseling services. Another chatbot feature named Tay(a crowdsourced Microsoft-backed chatbot on Twitter) is another example of a chatbot feature that inflicted harm on users. It spewed vitriolic messages targeting Jewish people and other marginalized populations with little to no human involvement in the roll-out of this program. Rightfully, Twitter discontinued the chatbot feature after 24 hours.

Artificial intelligence (AI) is already making headlines from a policy and legal perspective. In 2023, the Biden-Harris administration passed the Executive Order on safe, secure, and transparent Artificial Intelligence (AI). It establishes robust AI security and safety guidelines and protects American citizens' privacy. The Executive Order also adopts American ideals and values, such as worker protection, leadership, innovation, equity, civil rights, and human rights. New York State is the only U.S. state in 2021 that has passed and enacted legislation that mandates employers to conduct regular bias audits for AI-backed tools for making employment decisions. In layperson's terms, companies must inform prospective candidates that they use artificial intelligence in hiring and promotion efforts. Twenty-four other U.S. states have proposed or passed legislation to regulate the use of AI, protect users' data privacy, and mandate regular disclosures regarding how AI is used and how users' data are protected. No existing federal legislation is officially legally enshrined in the United States. However, the American Data Protection and Privacy Act (ADPA) is a U.S. federal privacy bill that was proposed but stalled during the previous Congressional session. The proposed privacy bill would set out to operationally define best practices and standards for AI and establish thresholds for potential risks that would directly impact companies and their roll-out of AI technological products. The European Union is in the process of passing its first comprehensive AI act. In late January of this year, members of the European Parliament released the final draft of the act to the general public.

The field of AI is vast and full of limitless potential to change the trajectory of our lives for the better if we harness it ethically and responsibly.

As we reach the end of this exploration, let us reflect on the double-edged sword power of artificial intelligence (AI). It can alter our lives for better and for worse. The vast, limitless potential of AI cuts across various sectors. It holds the promise to enhance individual well-being and societal health. My interaction with AI tools, namely the Summit AI app, has been extremely pivotal in helping me to maintain focus on achieving my goals, which exemplifies AI's power in serving as a supportive companion in our personal development and growth. While we may look at the innovation of AI with rose-colored glasses, it is essential to acknowledge and be mindful of the potential issues that AI can bring, namely biases, which can mislead us and cause disadvantaged groups in society to endure discrimination and exclusion based on of erroneous data or the unchecked biases that we as humans innately have. The path to AI innovation requires us to be optimistic and adopt a realistic approach. By harnessing AI responsibly and ethically, we can empower ourselves to strive toward achieving our personal goals and cultivating a more just, equal society one day at a time.

Keywords: artificial intelligence, machine learning, technology, computer science, neural network, European Union, comprehensive AI Act, federal legislation