What is Artificial Intelligence (AI)?

Artificial intelligence (AI) is a technology that processes information and learns from examples and patterns rather than just following exact instructions.

"AI is when computers learn to solve problems and make decisions, similar to how humans use their brains to understand the world around them,” said David Humphreys, the director of AI integration at Southern New Hampshire University (SNHU). “Instead of just following instructions step-by-step, AI learns by looking at examples, like recognizing your voice to play your favorite music, suggesting shows you might enjoy or helping self-driving cars know when to stop or turn."

What is AI in Simple Terms?

AI is often described as a computer program that tries to think the way people do — processing information, spotting patterns and making decisions based on examples rather than instructions. “It’s a system that mimics how humans reason, but does so in ways that are faster, sometimes broader and often surprising,” said Humphreys, who has 10+ years of experience working at the intersection of education, technology and media.

Traditional AI is built to perform specific tasks like recommending TV shows or flagging spam emails. It works by learning from large datasets to recognize patterns and make predictions, he said. For example, if you show it thousands of photos of cats, Humphreys said, it can learn to spot what a cat looks like.

But it doesn’t actually understand anything — it just finds patterns and makes educated guesses based on what it has seen before, he said.

Types of AI

While AI has become a widely used term, there are actually many different types of artificial intelligence. Here are a few, according to Phillip Peng, an SNHU director of data science and scaling AI:

• Computer vision enables machines to interpret and understand visual information from images and videos. This technology has use cases in areas like facial recognition, medical imaging and autonomous vehicles.
• Generative AI goes a step further than traditional AI by creating new content rather than simply analyzing existing data. These systems can produce text, images, audio or even code based on the patterns they've learned. Examples include large language models like ChatGPT and image generation tools like DALL-E.
• Natural language processing (NLP) focuses on understanding, interpreting and generating human language. This includes sentiment analysis, translation, text summarization, chatbots and speech analytics.
• Predictive analytics/machine learning uses historical data to identify patterns and make predictions about future outcomes, such as enrollment forecasting, customer behavior prediction and risk assessment.
• Recommendation systems provide personalized suggestions based on user preferences and behaviors, commonly used in e-commerce, streaming services and educational platforms.
• Reinforcement learning is trained through trial and error by receiving either rewards or penalties. This is used in robotics, game playing and optimization problems.

How Does AI Work?

AI generates responses through a method similar to autocomplete, but far more advanced. "It breaks user input into smaller parts called tokens and predicts what comes next, much like how your phone suggests the next word in a text message, but on a larger, more complex scale," Humphreys said.

This process allows AI to generate responses by analyzing patterns rather than understanding meaning.

"For example, if you were to ask it, 'Describe today's weather in a literary way,' it might output, 'It's raining cats and dogs,'" he said. To generate this response, Humphreys said AI would start with the word "It's" as a token, then predict the next logical word, such as "raining," or another weather-related term.

"For each token it generates, the AI looks ahead and determines what is most likely to follow," he said. "It uses vast amounts of data to make these predictions, but the results aren’t always perfect."

Because AI operates on probability rather than comprehension, it can sometimes produce inaccurate or unexpected responses. "It might say 'It's raining meatballs,' in which case it's probably not correct," Humphreys said. This predictive nature allows AI to perform tasks, he said, but also explains why it can sometimes produce errors or unexpected outputs.

When AI provides incorrect or unrelated responses, these are known as hallucinations. According to Humphreys, hallucinations happen when an AI model produces information that is inaccurate or nonsensical. "This occurs when the model misinterprets data or generates details that were not part of the original input," he said.

What Are Examples of AI?

AI can be seen almost everywhere, from everyday tasks to highly specialized work. "AI is really all-encompassing if we're thinking about it at a high level,” Humphreys said.

It covers many different areas and includes technologies like machine learning, he noted, which has been a core part of AI systems for years, enabling computers to learn and improve from experience rather than just following set instructions. Humphreys said that common daily examples of AI use include virtual assistants that you might use every day, like Alexa, Google Home Assistant and Siri.

However, recent developments in AI, known as generative artificial intelligence, create entirely new kinds of output based on your requests. According to Humphreys, generative AI takes your input and uses patterns it has learned from large amounts of data to produce responses.

Examples include writing answers to your questions, creating voice recordings or making videos and images based on your descriptions, he said. Some popular AI tools include:

• Code Assistants: Codeo
• Digital Avatar Creation: HeyGen, Synthesia
• Image Generation: Midjourney, DALL-E, Stable Diffusion
• Large Language Models (LLMs): ChatGPT, Claude, Copilot, Gemini, Latimer
• Music Generation: Suno
• Video Generation and Editing: Kling, Sora
• Voice and Speech Synthesis: ElevenLabs

"It seems like every week there's a new tool that fills a need that previously AI didn't touch," Humphreys said, also noting that the list of AI examples grows so quickly that it's nearly impossible to keep it complete and up to date.

Is AI Good or Bad?

AI is not categorically good or bad. AI has its advantages, such as deftness in pattern recognition, automation of repetitive tasks and general speed of output. Strategic use of AI can have outcomes that help humans at the micro and macro levels.

But, it's also a technology that comes with multifaceted risks — ones that require careful consideration, according to Peng. He listed the following:

• Bias and fairness. Because AI relies on data input by humans, it can perpetuate and amplify existing societal biases present in training data, leading to discriminatory outcomes in areas like hiring, lending, criminal justice and education. Learn more about AI bias.
• Environmental impact. Training large AI models requires significant computational resources and energy consumption. Discover more about the environmental impact of AI.
• Ethical concerns. Issues around consent, surveillance, manipulation and the use of AI in sensitive domains — like education and healthcare — require careful ethical consideration.
• Job displacement. Automation through AI may disrupt certain job markets, requiring workforce retraining and adaptation.
• Lack of transparency and accountability. Complex AI systems can make decisions that are difficult to explain or challenge, raising questions about accountability when things go wrong.
• Over-reliance and deskilling. Excessive dependence on AI systems may lead to erosion of human skills and critical thinking abilities.
• Privacy concerns. AI systems often require large amounts of personal data, raising concerns about data collection, storage, usage and potential breaches.
• Security vulnerabilities. AI systems can be manipulated through adversarial attacks.

And, like many tools and technologies, humans can use AI for good or bad. An example of a benevolent application might be AI's use in medical imaging for earlier disease detection. Meanwhile, generative AI can be used maliciously to generate deepfakes and spread misinformation. That's what makes understanding AI ethics so important.

"With AI and similar tools that are especially vulnerable to biases in data collection, gathering and output, it's vital that users remain ethical and vigilant in their use," Humphreys said.

Can AI Replace Human Intelligence?

No, AI cannot fully replace human intelligence, and there are a few reasons for that.

According to Peng, who has more than 18 years of experience driving AI and machine learning solutions, one of the most persistent misconceptions people have about AI is that it possesses human-like understanding and consciousness.

"Many people believe AI 'thinks' or 'understands' in the way humans do," he said. "In reality, even the most advanced AI systems are sophisticated pattern recognition tools operating on statistical correlations."

Simply put, AI is a tool with limitations — ones that require human input or intervention. For example, Peng said AI cannot:

• Adapt to completely novel situations. AI performs poorly when encountering scenarios significantly different from its training data.
• Be truly creative. While AI can generate novel combinations, it’s limited by its training data and doesn’t possess genuine creativity or innovation.
• Exercise common sense. AI struggles with reasoning that humans find intuitive.
• Explain its reasoning fully. Many AI systems struggle to provide transparent explanations for their decisions.
• Feel emotions or empathy. AI simulates emotional responses but doesn’t experience feelings.
• Make ethical judgments. AI lacks moral reasoning and cannot navigate complex ethical dilemmas independently.
• Understand context deeply. AI lacks true comprehension of meaning, culture and nuanced human experience.

And other experts agree. "The limitation of AI is that it lacks true understanding, emotional intelligence and adaptability," said Maggie Aubin, an AI integration specialist at SNHU. "It depends heavily on data quality and human oversight, and it cannot replace human judgment in complex, ethical or ambiguous situations."

According to Humphreys, creativity is uniquely human. And while AI tries to copy creativity, it can't fully capture it. "If you ask it for a poem, it will write a poem,” he said. “Is it going to be a good poem? Probably not compared to a poet laureate or somebody who has written poetry."

AI might seem very creative if it's working in an area where you're not as skilled. But it's still only imitating human creativity, not truly being creative on its own, he said.

What Can AI Do Better Than Humans?

AI isn't necessarily smarter than humans, but it can do certain things much faster. "The most effective applications of AI combine machine efficiency with human wisdom," Peng said, offering the following use cases:

• Automation. Performing repetitive tasks at scale with consistency and speed.
• Content generation. Creating text, images, code and other content based on learned patterns.
• Forecasting. Predicting outcomes based on historical data.
• Pattern recognition. Identifying complex patterns in large datasets that humans might miss.
• Personalization. Tailoring experiences and recommendations based on individual preferences.
• Processing speed. Analyzing vast amounts of data quickly.

Humphreys had similar thoughts on that last point. "The big thing that AI has that humans don't is speed," he said. "You could ask it to output an entire piece of software, and it could do that in a short amount of time. Would the software run? Would it be coded well? Those are two completely different questions, but it would output code very quickly."

AI can handle large tasks in seconds or minutes that might take you hours or days. And according to Humphreys, that means AI can be great for repetitive or tedious tasks — like summarizing long meetings, drafting emails or proofreading writing.

Work done by humans will continue to be necessary, though. "The future isn’t about AI replacing humans," Peng said, "but about creating synergies where AI handles routine tasks and data processing while humans focus on critical thinking, creativity and interpersonal connections."

Read more about the future of AI.

What Are the Real-World Uses of AI?

AI can be best used as an assistant, helping you with tasks that are repetitive or take a lot of time.

"The best way most people can use AI is thinking about it as an assistant, having it work on things that are time-consuming or that you don't like to do,” Humphreys said. “By using AI in that way, you're not undermining your own creativity."

For example, Aubin said you could use it to generate recipe ideas, personalize workout routines or plan your next vacation. At work, she suggested using it to help check the tone of your emails, plan your workday or build a project structure.

Using AI to organize your thoughts can help you clearly outline ideas, spot gaps and save time planning or structuring your work. “I'll turn on voice mode on ChatGPT, and I'll talk to it,” Humphreys said. “And then at the end, I'll just ask it to create a summary of the ideas I had or to help me flesh out an idea.”

Similarly, Aubin uses AI to turn large lists of to-do tasks into actionable and bite-sized activities — something she said helps her as a neurodivergent person. "I can get overwhelmed if I have too many directions to follow without a clear plan," she said. "So it helps to simplify things."

So, how do you get into artificial intelligence? For Aubin, it was curiosity. Coming from a liberal arts background with experience in academic advising, she really wanted to understand how AI works and the impact it would have on students. Ultimately, that led to her transition to SNHU's AI team as an integration specialist.

If you're interested in learning or building a career in AI, you can start by trying out and researching AI models. "Play around with them,” Humphreys said. “Figure out the ones that connect with you, whether that's (with) your ethics, your values or just the output of the model."

Aubin said you could also look for free training online from tech companies like LinkedIn Learning, Google and IBM. But she stressed the importance of finding courses specific to your industry, as that's how you'll learn applications of AI that are relevant to you.

If you want to take your learning further, you can earn real AI credentials — like badges and certificates — online. In fact, SNHU has a 4-course online generative AI concentration you can add to a variety of bachelor's programs. You can also receive a Generative AI Practitioner badge by successfully passing a hands-on, 6-week course.

Read more: How to Learn Artificial Intelligence (Plus Helpful Courses and Skills)

Commonly Asked Questions About AI

Here are a few questions you might have about artificial intelligence.

Are Some AI Models More Ethical Than Others?

Yes, different AI models prioritize ethics in different ways, such as reducing bias or improving transparency, said Humphreys. "(Claude AI) has constitutional AI, which is a way of defining how the AI interacts with information for the user and within the kind of experience itself."

Latimer is an AI model based on an older version of ChatGPT, specifically trained to promote cultural inclusivity, with a focus on representing the Black experience, he said.

While some models, like DeepSeek, raise concerns due to how they handle and share information, Humphreys said.

What is Machine Learning?

Machine learning (ML) is a foundational approach in artificial intelligence that allows computers to learn from data and make decisions. It’s like teaching a computer to recognize patterns by giving it examples, he said. "Think of it like teaching a dog tricks with treats: the dog starts recognizing what behaviors get rewards."

What is Deep Learning?

Deep learning is a subset of machine learning that uses multi-layered neural networks to solve more complex problems. “It’s like the overachiever of the ML family,” Humphreys said. Instead of being told what features to look for, deep learning models figure it out on their own — especially useful for tasks like recognizing faces or generating text, he said.

What is Artificial Narrow Intelligence?

Artificial narrow intelligence (ANI) refers to the kind of AI we use today, designed to perform well at a limited set of tasks. It’s effective at things like generating text, creating images or writing code — tools like ChatGPT and Google Gemini are examples, Humphreys said. But for any task these systems can handle, there are still people who can do it better, he said.

What is Artificial General Intelligence?

Artificial general intelligence, or AGI, is a hypothetical form of AI that would outperform humans at nearly every task. “That would range from driving cars to math to programming to whatever,” Humphreys said.

While AGI doesn’t exist yet, many believe it could arrive within the next five to 10 years, though we don’t know how big a disruption it’s going to be, how expensive it’s going to be to use, or who’s going to have access to it if it arrives at all, he said.

Ollie Burkett '25G is a writer with a Master of Arts in English and Creative Writing from Southern New Hampshire University. You can connect with him on LinkedIn.