What Is AI?

The future is here. (DALL-E generated image.)

AI, or Artificial Intelligence, is a technology that’s developing at pace, with increasing application in many fields. Its march sparks optimism and concern in equal measure. But what, exactly, is it?

In this article, we’ll look beyond the hype to explore how AI works, the threats and opportunities it presents, and how you can make smart use of this rapidly evolving tool.

1. What AI Means

AI programs can perform tasks traditionally carried out by humans. These could include holding conversations, recommending products, or recognizing patterns in images, for example. [1]

AI may stand for “Artificial Intelligence,” but that doesn’t mean that AI programs think the same way a human does. AI makes decisions based on the data it's being fed and can process huge amounts of information far more quickly and efficiently than any human brain.

2. What AI Doesn’t Mean

Sometimes, “AI” is used as a buzzword when referring to technologies that don’t fit the definition. AI programs are designed to learn and adapt. Not all “clever-looking” technology can do these things – and if it can’t, like the examples below, it doesn’t qualify as AI.

• Some chatbots are AI, but not all. Some follow scripts and can’t adapt or generate original responses.
• Most robots are not AI at this point. Industrial robots, for example, might carry out repetitive, movement-based tasks without learning or making decisions.
• Image, speech recognition, or predictive models (like forecasting tools) might look advanced but rely strictly on predefined rules.
• Supercomputers have a very high level of performance and can do extremely fast calculations. But they aren’t necessarily AI.

3. The Origins of AI

The concept of Artificial Intelligence has been around at least since Ancient Greece, where philosophers and playwrights popularized myths about mechanical men (such as the bronze warrior giant Talos created to guard the island of Crete). [2]

The term “Artificial Intelligence” was coined by computer scientist John McCarthy at the Dartmouth Conference of 1956. A couple of years later, the first artificial neural network (a type of AI that mimics how the human brain works), “Perceptron Mark I,” was built. [3]

Modern AI programs can now recognize images and language, produce text and media output, play games, drive vehicles, diagnose diseases, and more.

In 2022, AI research nonprofit OpenAI released the large language model, ChatGPT. It can respond to users’ prompts and instructions, producing text-based answers to questions, or writing documents.

Despite modest expectations, ChatGPT became one of the most popular apps ever launched. [4]

4. Types of AI

Broadly speaking, there are three different types of AI:

1. Weak/Narrow AI: This is the only AI that exists currently. It focuses on carrying out a single task – and can often perform that task faster and better than people. These technologies have no genuine intelligence; rather, they do what they were designed to do. In Siri's case, it processes our queries and searches for results. ChatGPT, for example, conducts text-based chat.

2. Strong/AGI (Artificial General Intelligence): In theory, this AI would perform any intellectual task that a human can, including reasoning, problem solving, planning, and decision making. Strong AI models would be able to use previous learning to carry out new tasks without first being trained by humans.

3. Superintelligence: This type of AI would be able to reason, learn, and think. Its capabilities would far surpass the cleverest human brains. But it’s unlikely to exist until the distant future, if at all. [5]

5. Functions of AI

AI can be grouped into categories based on how it functions.

• Reactive Machine AI has no memory. It performs a specific task by looking at the data provided and analyzing it using statistics. IBM Deep Blue, the chess-playing computer, is an example.
• Limited Memory AI can use past and present data to decide on the best option for achieving an outcome, which means that its performance improves over time. A generative AI model like ChatGPT uses limited memory to predict the next word that fits best within a context. [5]

AI has resulted in advances across many fields, from agriculture to medicine to customer service. Most of these have involved a type of AI called machine learning, which uses algorithms – sets of rules and instructions – that are trained on large sets of data until the AI is able to make choices based on experience.

Subtypes of machine learning include deep learning. This uses neural networks with many layers to process data and predict results, and Natural Language Processing (NLP), which can analyze and generate language-based communication. [6]

6. AI Threats

Key challenges presented by AI include:

• Job displacement: According to a report from ResumeBuilder, 37 percent of U.S. business leaders say that AI replaced human employees in 2023. Many of those jobs were lower level or easier to automate. This doesn’t mean AI will “take your job.” It’s more likely jobs will go to people who know how to use the latest AI technology.
• Inaccuracy and hallucination: AI can produce inaccurate information and judgments. Computer vision might mistake a poisonous plant for an edible one. Or it could make up references, such as books or legal cases, that don’t exist. (When an AI invents information, it’s known as “hallucination.”)
• Built-in bias: AI will replicate biases embedded in the datasets used to train it. This might result, for example, in AI trained to sort through candidates for jobs recommending resumés with male or Western-sounding names.
• Safety concerns: Some AI uses personal data to, for example, recommend books or films. This raises questions of privacy if AI accesses too much. Serious security concerns could emerge if AI leads to cyberattacks or control over weapons.
• Ethical concerns: Lawsuits have emerged regarding violation of copyright law when generative AI uses material without the creators’ permission. Some countries have banned ChatGPT due to concerns related to privacy or misinformation.

7. AI Opportunities

AI offers many opportunities in the professional world. Here are just a few.

• Rapid productivity: Generative AI offers the ability to create things such as written documents, art and videos rapidly and to your specifications. AI in general can often perform tasks much faster, and in some cases better, than humans, which can increase efficiency.
• Cost savings and profitability: Businesses are already reporting significant time and cost savings due to AI, allowing them to redirect resources. This can vastly improve efficiency, helping to boost profitability, productivity and competitiveness, saving hundreds of billions of dollars annually. [7]
• Accessibility: People communicating in a non-native language can use AI to draft clear correspondence without grammatical mistakes. And anyone can learn to prompt an AI model to create an image.
• Quick analysis: AI can process large amounts of data and generate insights with amazing speed.
• Personalization: Many AI tools can be tailored to the user, whether that’s meeting the needs of an individual customer or creating a virtual assistant that can keep you organized. This trend is set to continue, with AI becoming more and more accessible and customizable.
• Innovation: AI has powered advances   in many fields. In healthcare, it’s being used to develop new, personalized treatments. It’s also behind the development of self-driving vehicles and has contributed to new scientific understandings.

Early adopters of the latest AI technologies will be able to get more done, faster, and have the opportunity to move these tools in exciting new directions   .