The Top Three Pros and Cons of Artificial Intelligence

10 May 2024

Artificial intelligence (AI) was once relegated to science fiction. Today, multiple businesses are leveraging AI to realize their goals. Technology companies see AI as a critical area to set themselves apart from competitors, while world governments explore regulating it to protect their citizens. No matter where you look, everyone is talking about AI.

While AI gets a lot of press, most coverage centers around how it's revolutionizing the economy or jeopardizing humanity. With no discussion of what artificial intelligence is or how it works, people need help to maintain a rational discourse on the subject.

This article strives to change that. It won't try to convince you that AI is the "next big thing" or an existential threat to society. Instead, it'll use layman's terms to explain what AI is, shedding light on three of its biggest pros and cons. By the end, you should be able to formulate an opinion on AI driven by the facts, not hype and misinformation. First, AI must be defined.

What Is Gen-AI?

Artificial intelligence (AI) may be defined as a machine's ability to learn, reason, generalize, and infer meaning. It is a very broad term that allows many things to be categorized as AI: video games, chess bots, search engine algorithms, virtual assistants, machine learning, text generation, and more. Most of these applications have been around for decades and aren't responsible for the current interest in AI.

When the media, business leaders, or the government talk about AI today, they typically talk about generative AI or gen-AI for short. Gen-AI is defined as AI producing text, images, music, or art in response to a prompt. Gen-AI is built using large language models (or LLMs) and allows a machine to use algorithms, pattern recognition, and complex processes called neural networks to simulate the human decision-making process. LLMs are "trained" on the web to get an initial database to work with and improve as users ask them to complete tasks.

ChatGPT is the most famous example of gen-AI and is sometimes used synonymously with it, though it really shouldn't be. Users can ask ChatGPT to compose an outline for written assignments, write simple stories, or answer factual questions. The AI will deliver a text-based reply almost instantaneously based on what it's "learned" from the internet and prior use.

ChatGPT and comparable programs need help to understand what they're saying. For example, AI frequently starts a story with "once upon a time" because many of the stories it was trained on started that way. It doesn't know why so many stories begin with that phrase, but it recognizes a clear pattern in that they do. Most of the doomsday scenarios involve AI becoming self-aware and thinking like a human, but modern gen-AI is nowhere near this type of general AI.

Instead, AI can only complete particular tasks defined by their programming. Put another way, ChatGPT can only do something once a human gives it a text-based prompt.

3 Pros of Gen-AI

  1. Gen-AI Promotes Workplace Efficiency

    One of the biggest benefits of gen-AI is automation. You've seen this in action if you've ever turned to a phone recording or chatbot to resolve a customer service request. The AI is trained on a company's policies and delivers that info to customers more quickly than a human could. This reduces the company's labor costs while giving customers a streamlined service experience.

    Other fields can also benefit from gen-AI. For instance, teachers can use gen-AI to create lesson plans that are compliant with state standards, corporations can use it to streamline the onboarding process, and healthcare professionals can use it to automate aspects of medical screenings. The possibilities are limitless.

  2. Gen-AI Jumpstarts Creativity

    Similarly, creatives can use gen-AI to expedite content creation. A writer might use it to outline a project quickly, perform keyword research in a fraction of the time it would take by hand, and double-check their work for typos. An artist could turn to AI-generated images for inspiration or a sketch to get them started.

    Gen-AI is also a powerful research tool whether a fifth-grader is building a diorama of an arctic ecosystem or a lawyer is preparing for a case. Research is traditionally a time-consuming endeavor, and streamlining it can dramatically improve productivity.

  3. Gen-AI is Fun and Convenient

    While Gen-AI has many practical applications, many people find it fun to play around with. There are AI tools for generating absurd images, writing ridiculous stories, and even playing interactive games that can respond to a player's actions.

    Likewise, Gen-AI makes information more accessible in social situations. Want to settle a bet with a friend? ChatGPT can find the answer much more quickly than you could otherwise. Gen-AI will also tell you what you want to know as soon as you want, even if you lack the time or inclination to research it.

3 Cons of Gen-AI

  1. Gen-AI has Hallucinations

    Gen-AI only sometimes answers questions correctly, creating problems when users blindly trust it. The AI industry describes these errors as "hallucinations." In one famous example, lawyer Steven Schwartz used ChatGPT to expedite his research and filed a legal briefing citing cases that never happened. He lost his case and was charged a $5,000 fine for falsifying evidence.

    AI bias is also a significant issue. Since all LLMs are trained on the internet, they find a variety of untrustworthy information. Since AI tools don't understand what they're reading, they cannot distinguish between reliable sources and conspiracy theories. Therefore, they frequently cite nonsense as fact and lend credibility to erroneous info.

  2. Gen-AI is Expensive

    Most people assume gen-AI is free because ChatGPT has a free version, but that's not true. ChatGPT was invented by a company called OpenAI, and their free version is nothing more than a marketing tool. If you want the best ChatGPT has to offer, you need to pay for GPT4 or ChatGPT Enterprise.

    Gen-AI pricing needs to be more counterintuitive. Users are charged fractions of a cent per token, but tokens are arbitrary. For instance, the word "hamburger" is three tokens by itself. The costs add up fast, so many applications must be more flexible about how many tokens can be used per query. Users waste tokens on unoptimized queries because they need help understanding how it works.

  3. Some People Hate Gen-AI

    Many people fear AI and are taking practical steps to minimize its effectiveness. For example, AI is trained by crawling the web. Artists who feel that Gen AI jeopardizes their livelihood have started data poisoning their work, adding code to digital images that sabotage Gen AI's ability to learn from them. These "sleeper agents" are hidden until the AI fails to produce an image.

    Companies have also faced intense public backlash from trying to use gen-AI to streamline projects. For instance, Wizards of the Coast (the company behind the popular role-playing game Dungeons and Dragons) faced boycotts once it was discovered that AI was used to generate art in some of its books.

Generative AI likely isn't going anywhere, but its capabilities are different from what its biggest components and detractors claim. You still need a human touch if you hope to get the most out of it.