Intelligent Machines and Where We Are in the AI Journey
AI 101: Article 1 “What is Artificial Intelligence?” by Bart Niedner, 26 June 2023
What Is AI?
Artificial Intelligence (AI), often shrouded in mystery and hype, has become a buzzword that permeates our technological landscape. What does it mean? With so many technical terms and acronyms, getting lost in the maze of jargon is easy. “AI” is often misunderstood. Let’s consider the definition of AI and approach this topic with intellectual curiosity, humility, and a desire to engage in the ongoing discussion.
“Intelligent Machines”
Brief History
1956 Origins
John McCarthy, along with Marvin Minsky, Nathaniel Rochester, and Claude E. Shannon, first coined “Artificial Intelligence” in a paper for the 1956 Dartmouth Summer Conference. They defined it as “the science and engineering of making intelligent machines.” Unfortunately, this elegant, straightforward definition shifts much of the meaning to the nebulous question, “What is intelligence?” Most modern definitions for AI focus the litmus test on human intelligence and provide examples to clarify this elusive and nuanced phrase.
AI’s Golden Era
DARPA (Defense Advanced Research Projects Agency) heavily funded AI research through the mid-70s. The hype in the mid-50s caused major investment in AI research with few requirements for developing anything functional. Though the technology had many limitations, significant progress was made, particularly in the areas of expert systems and natural language processing.
AI Winter
The term “AI winter” was coined in the mid-70s as funding for AI research decreased due to a lack of significant progress and unrealistic expectations. However, despite these challenges, the groundwork for future advancements in AI was being laid, and the field continued to grow and evolve in the following decades.
Our Long AI Summer
The AI winter ended in the early 1990s due in part to the emergence of new techniques and technologies. The explosion in distributed computing power and access to vast data that began in the 1990s allowed for significant breakthroughs in AI research. These breakthroughs, in turn, continue to spur more investment.
Speaking of spurs, John Wayne passed in 1979 during the AI winter. I wonder how The Duke would describe what constitutes AI if he were alive today. Wouldn’t it be cool if we could ask AI what he might say? There are AI tools that can approximate celebrity voices in an audio format. But for our blog purposes, let’s just ask a chatbot to do it in text.
Let’s prompt ChatGPT and see what we get: “Please use an analogy, include a non-technical explanation, and add a real-life example to explain what makes something qualify as ‘artificial intelligence.’ Do it in John Wayne’s voice.”
Well, partner, picture this. You’re out on the plains, riding under the wide-open sky. You’ve got a trusty sidekick, a smart horse that you’ve trained to respond to your voice. When you say, ‘giddy up,’ she gallops. When you say, ‘whoa,’ she slows down. She even understands when you want her to follow a path or avoid obstacles. Now, that’s a well-trained horse, but in the end, she’s following your commands — she ain’t making her own decisions.
— ChatGPT‑4 (in the voice of John Wayne, d. June 11, 1979)
Now, suppose you had a horse that didn’t just follow your commands, but could also make decisions on her own, like figuring out the best path to take to avoid outlaws or finding a water source when she’s thirsty. Now, that would be somethin’, wouldn’t it? That’s the difference between a regular old program and artificial intelligence.
In the world of technology, artificial intelligence, or AI, is like that second horse. It’s not just following commands. Instead, it’s designed to do tasks that normally require human thinkin’. Things like understanding what folks say, recognizing patterns, learnin’ from what happens, and making decisions.
That’s what makes something qualify as ‘artificial intelligence,’ pardner. It’s not just about following commands — it’s about thinkin’, learnin’, and makin’ decisions, just like a human would. Now, that’s one smart horse!
That’s a pretty good example of AI. Nobody programmed the question or answer into the system. ChatGPT used its training experiences to respond to a nuanced question and responded with a pretty convincing answer in a novel voice.
The Common Definition of AI
Artificial Intelligence is the development of computer systems capable of performing tasks that typically require human intelligence, such as understanding natural language, recognizing objects, learning from experience, and making decisions.
What AI Is Not
Considering what artificial intelligence is not can help clarify our understanding of AI. For example:
- Not all advanced technology or automation qualifies as AI. A traditional scientific calculator or the traditional Google search engine algorithm doesn’t possess the learning or reasoning capabilities associated with AI.
- Traditional computer programs are not AI. Conventional software programs follow predefined rules and instructions without adaptive or learning abilities.
- Automation is not AI. Automated systems perform predefined tasks without exhibiting intelligence or adapting to new situations.
- Statistical analysis is not AI. Statistical techniques and data analysis can be powerful tools but do not inherently constitute AI because they rely on predetermined methods which do not change.
It is also important to recognize that most advanced technology exceeds human intelligence’s capacity in specific areas — it is the underlying nature of all tools. However, the incredible computational power in advanced technology alone should not be conflated with AI.
Where are we in the AI Journey?
To better grasp our current stage of AI development, it is helpful to understand the differences between weak AI, strong AI, and Super AI. These three stages are commonly used to describe the levels of technological development for AI.
Weak AI (ANI)
Weak AI is also called Narrow AI because it is designed to perform specific or limited tasks. Examples include recommendation algorithms on streaming platforms like Netflix or Spotify DJ, voice assistants like Siri or Alexa, and even autonomous vehicles. These systems excel in their designated functions but lack versatility for tasks outside their scope. Narrow AI is limited to learning and adapting in a specific arena. For example, a Narrow AI chatbot such as ChatGPT may excel, learn, and adapt in understanding natural language but misunderstand the context of that conversation. A Narrow AI image generator, such as Midjourney, may produce an image that is indistinguishable from a real-life digital photo. However, its inability to understand the concept of depth is a challenge when prompting.
Strong AI (AGI)
Strong AI is also commonly called General AI. In contrast to Weak AI, Strong AI refers to a hypothetical machine capable of understanding and performing any human intellectual task. This type of AI remains the subject of ongoing research and has yet to be achieved. Strong AI would possess human-like cognitive abilities, allowing it to learn, understand, and adapt across all human tasks.
Super AI (ASI)
Super AI, also called Artificial Super Intelligence, refers to an advanced form of artificial intelligence that surpasses the capabilities of human intelligence in various domains, such as problem-solving, decision-making, and learning. It can process and analyze vast amounts of data at a speed far exceeding human capacity and perform complex tasks with high accuracy and efficiency. Super AI is still a theoretical concept, and its development and implications are subject to ongoing research and debate.
The Current Stage of AI Technology
The recent explosion of AI tools into the consumer market is a big deal precisely because they represent an impressive step from weak AI toward strong AI. Using ChatGPT or DALL‑E for the first time feels very close to strong AI. But extended use reveals these technologies as weak AI because they do not fully understand the real-world context. We are still firmly in the age of weak AI, but the line of distinction for strong AI is becoming much fuzzier as we approach it. We are on the exciting (or scary?) verge of machines with human-like cognitive abilities.
Push on the AI Definition
Definitions and facts are important. They are the building blocks for our understanding. But encouraging engagement in the AI conversation is the purpose of this article and blog. So, let’s not only explore the AI landscape; let’s be a part of creating it! Let’s push on the commonly accepted definition of AI and see what we find!
Note the qualifiers of learning and adapting in the “What AI is Not” section’s examples (above). The consistent presence of these qualifiers creates a strong argument for the necessity to include adaptability in the definition of AI if we want to compare this technology to human intelligence.
What Constitutes “Human Intelligence?”
At a 1921 symposium, psychologist Lewis Terman convincingly argued that a core quality in defining “human intelligence” is the ability to adapt. In this context, adaptation means changing oneself to cope more effectively with the environment, but it can also mean changing the environment or finding an entirely new one. The ability to abstractly and creatively manifest something new is a hallmark of successful adaptation. AI must include the characteristic of adaptability to emulate human intelligence convincingly. Robust study since 1921 has provided exciting and compelling breadth and depth to the concept of human intelligence. Various “theories of intelligence” have emerged, including psychological measurement, cognitive psychology, cognitivism-contextualism, and biological paradigms have given us parallax on the nuances inherent in human intelligence. Emotional and creative intelligence seem particularly fruitful areas of discussion for AI development. But there is a unifying quality – these theories consistently uphold the core notion of creative adaptability.
Creative Adaptation
The common definition of Artificial Intelligence could be better if it brought forward the unifying and prominent distinction for human intelligence: creative adaptability. Creative adaptability is broadly contained in terms such as “learning from experience” and “making decisions”; however, this fundamental distinction should be more prominently stated as a requirement. This point is particularly relevant when considering that the quality of a system’s adaptability is a major delineator moving into Strong AI. But across the board, it can be argued that if a computer system cannot creatively adapt, it is not artificial intelligence. For example, a calculator does not creatively adapt. It is procedurally programmed to respond and cannot be trained to perform more desirably — it can’t get smarter. In contrast, AI chatbots can respond to a broad context, and they use the inputs they receive from users to become better at what they do (as well as being trained using large data sets to improve their performance and accuracy).
An Improved Definition of AI
Artificial Intelligence is the development of computer systems capable of creative adaptation and performing tasks that typically require human intelligence, such as understanding natural language, recognizing objects, learning from experience, and making decisions.
What do you think? Is this a useful improvement?
Your Role in The AI Discussion
Please engage in the ongoing AI discussion here and elsewhere. It is essential to remain informed, curious, and open to its potential. Let’s explore the possibilities of AI together! What are your thoughts? What would you like to explore?
AI 101
AI-RISE articles in the AI 101 series are introductory material for anyone who wants accurate, conversational knowledge of this important technology shaping our world. This article makes the discussion more accessible to someone new to the AI conversation.
Article by Bart Niedner
All hail our technological overlords!
— Bart Niedner
Now, where did I put my eyeglasses?!
Bart Niedner, a versatile creative, embarks on a journey of discovery as he delves into both novel writing and the intriguing realm of AI-assisted writing. Bart warmly welcomes you on this journey from novice to master as he leverages his creative abilities in these innovative domains. His contributions to AI-RISE and BioDigital Novels reflect AI collaboration and exploratory work – the purpose of these websites.
“Get Your Geek On!” (Related Reads)
- Wikipedia contributors. “John McCarthy (Computer Scientist).” Wikipedia, June 2023, en.wikipedia.org/wiki/John_McCarthy_(computer_scientist).
- Sternberg, Robert J. “Human Intelligence | Definition, Types, Test, Theories, and Facts.” Encyclopedia Britannica, 19 July 1998, www.britannica.com/science/human-intelligence-psychology.
- Lateef, Zulaikha. “Types of Artificial Intelligence You Should Know.” Edureka, May 2023, www.edureka.co/blog/types-of-artificial-intelligence.
- Wikipedia contributors. “Technological Singularity.” Wikipedia, June 2023, en.wikipedia.org/wiki/Technological_singularity.
- Lutkevich, Ben. “AI Winter.” Enterprise AI, Aug. 2022, www.techtarget.com/searchenterpriseai/definition/AI-winter?Offer=abMeterCharCount_ctrl.
Encourage Participation
Interested?
Featured Image
Image Creation Remarks
This featured image was created with Midjourney and Photoshop v24.6.0 Beta. The transition from greyscale at the bottom to the inclusion of gold elements and finally to the blue energy at the top represents the transitions from conventional computing to narrow AI to General AI. The diagonal line represents where we are today, and the box it forms below indicates the confinement of narrow AI’s ability to adapt. The diagonal reflects that we are on the precipice of AI “seeing” beyond this container in our pursuit of GAI.
I used Midjourney to create the base image, which uniformly had the gold elements and none of the biological shapes, hair, or blue synapse energy. I used Photoshop’s beta release of their “Generative Fill tool” to add the biological shapes, hair, and blue synapse energy in a single pass. This was my first use of the Generative Fill tool. The ability to natively use AI image generation in Photoshop and, more importantly, in a precise, context-aware area of my workflow is game-changing! The gradient in the background is original to the Midjourney product. The grayscaling of the subject was a desaturated layer and layer mask. The border box was also a simple Photoshop artifact.
I used Midjourney to create the base image, which uniformly had the gold elements and none of the biological shapes, hair, or blue synapse energy. I used Photoshop’s beta release of their “Generative Fill tool” to add the biological shapes, hair, and blue synapse energy in a single pass. This was my first use of the Generative Fill tool. The ability to natively use AI image generation in Photoshop and, more importantly, in a precise, context-aware area of your work is game-changing! The gradient in the background is original to the Midjourney product. The grayscaling of the subject was a desaturated layer and layer mask. The border box was also a simple Photoshop artifact.
Midjorney Prompt (base image)
“AI Unveiled, An Exploration of Artificial Intelligence’s Definitions, Distinctions, and Debates, Discover the true meaning of Artificial Intelligence in this comprehensive exploration of AI’s definitions, distinctions, and debates. Unravel the complexities of narrow and general AI, and join the conversation on AI’s role in our future.”
Photoshop Prompt (Generative Fill)
Sadly, I was so excited about exploring the new Generative Fill tool that I failed to record the prompt. The image I chose was the sixth option (the tool generates sets of three options).
Postprocessing
No postprocessing (unless you want to count using Photoshop to remove the upper-background for the PNG).