I have just researched about Artificial Intelligence, who started AI and the gradual improvement done over the years.

AI Foundation – Who, When and What?

1. Early Foundations (1940s-1950s):

The conceptual foundations of AI can be traced back to the 1940s and 1950s. In 1943, Warren McCulloch and Walter Pitts proposed a model of artificial neurons, laying the groundwork for neural networks(V, 2024).

The term “Artificial Intelligence“ was officially coined in 1956 at a conference at Dartmouth College. This conference, organized by John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon, is widely considered the birth of AI as a field of study(Elamin, 2024; V, 2024).

2. The Golden Years (1956-1974):

This period saw great optimism and progress in AI research. Key developments included:

• The development of the Logic Theorist by Allen Newell and Herbert A. Simon in 1955, considered the first AI program(Al-Amin et al., 2024).
• The creation of ELIZA in the 1960s, one of the first chatbots, by Joseph Weizenbaum(Al-Amin et al., 2024).
• The development of expert systems in the 1970s, which attempted to encode human expertise into rule-based systems(Elamin, 2024).

AI Research Ups & Downs

3. The First AI Winter (1974-1980):

Enthusiasm waned as early promises failed to materialize, leading to reduced funding and interest in AI research(Elamin, 2024).

4. Expert Systems Boom (1980-1987)

The development of expert systems brought renewed interest and funding to AI(Elamin, 2024).

5. The Second AI Winter (1987-1993):

Another period of reduced funding and interest as expert systems failed to live up to expectations(Elamin, 2024).

6. The Rise of Machine Learning (1993-2011):

This period saw a shift from rule-based systems to data-driven approaches:

• The development of statistical learning methods and neural networks gained traction(Elamin, 2024).
• In 1997, IBM's Deep Blue defeated world chess champion Garry Kasparov, marking a significant milestone in AI's capabilities(Al-Amin et al., 2024).

Where Are We Now?

7. Deep Learning Revolution (2011-present):

The current era of AI is characterized by breakthroughs in deep learning and neural networks:

• In 2011, IBM Watson won the quiz show Jeopardy!, showcasing advanced natural language processing capabilities(Al-Amin et al., 2024).
• In 2012, deep learning models achieved breakthrough performance in image recognition tasks(Elamin, 2024).
• The development of large language models like GPT (Generative Pre-trained Transformer) has led to significant advancements in natural language processing and generation(Al-Amin et al., 2024).

8. Current State and Future Directions:

Today, AI is being applied across various domains, including healthcare, finance, transportation, and education(Elamin, 2024; Sharma, 2020). Key areas of focus include:

• Explainable AI: Developing AI systems that can explain their decision-making processes(Elamin, 2024).
• Ethical AI: Addressing concerns about bias, privacy, and the societal impact of AI(Elamin, 2024).
• General AI: Working towards AI systems that can perform a wide range of tasks at human-level intelligence(V, 2024).

Throughout its history, AI has experienced cycles of high expectations followed by periods of disillusionment. However, each cycle has contributed to the field's overall progress(Elamin, 2024). The current era is marked by unprecedented advancements, but also growing concerns about the ethical implications and potential risks of AI technology(Elamin, 2024; V, 2024).

As we look to the future, interdisciplinary collaboration and a balanced approach to AI development will be crucial in realizing its full potential while addressing societal concerns(Elamin, 2024).

What is Artificial Neurons?

1. Definition and Concept:

Artificial neurons, also known as artificial nerve cells or nodes, are mathematical functions designed to mimic the basic functionality of biological neurons in the human brain (Kufel et al., 2023). They are the core processing units in artificial neural networks, receiving inputs, processing them, and producing outputs.

2. Structure and Components:

An artificial neuron typically consists of:

• Inputs: Representing the signals received from other neurons or external sources
• Weights: Numerical values that determine the strength of each input
• Summation function: Combines all weighted inputs
• Activation function: Determines the neuron's output based on the summation result
• Output: The final signal sent to other neurons or as the network's result

3. Types of Artificial Neurons:

There are various types of artificial neurons, each with specific characteristics:

a) Perceptron: The simplest form of artificial neuron, capable of binary classification.

b) Sigmoid Neuron: Uses a sigmoid activation function, allowing for smooth, non-linear outputs.

c) Rectified Linear Unit (ReLU): Employs a simple activation function that outputs the input if positive, otherwise zero.

d) Long Short-Term Memory (LSTM) units: Complex neurons designed for processing sequential data.

4. Activation Functions:

The choice of activation function significantly impacts the neuron's behavior and the network's overall performance. Common activation functions include:

• Sigmoid
• Hyperbolic tangent (tanh)
• ReLU and its variants (e.g., Leaky ReLU)
• Softmax (for multi-class classification)

5. Learning and Adaptation:

Artificial neurons can adapt their behavior through learning processes, primarily by adjusting their weights. This is typically done using optimization algorithms like gradient descent during the training phase of neural networks (Rathore et al., 2018).

6. Biological Inspiration:

While artificial neurons are inspired by biological neurons, they are significant simplifications. Recent research has shown that individual biological neurons may be more complex than previously thought, potentially capable of performing operations similar to multi-layer artificial neural networks (Beniaguev et al., 2019, pp. 2727-2739.e3).

7. Hardware Implementation:

Efforts are being made to create hardware-based artificial neurons for more efficient neuromorphic computing. For example:

• SQUID-based artificial neurons for superconducting quantum interference devices (Katayama et al., 2018)
• Memristor-based neurons for low-power, high-density implementations (Lu et al., 2020, pp. 1245–1248; Yan et al., 2023)
• Spintronic devices for energy-efficient neuromorphic computing (Yang et al., 2021, pp. 1–10)

8. Advanced Concepts:

Recent developments in artificial neurons include:

• Spiking neurons: More closely mimicking biological neurons by using discrete spikes instead of continuous values (Gu et al., 2024)
• Artificial complex neurons: Utilizing complex-valued weights and activations for certain applications (Kotsovsky et al., 2015, pp. 57–59)
• Oscillating neurons: Incorporating temporal dynamics into neuron behavior (Noel et al., 2021)

9. Applications:

Artificial neurons, as part of neural networks, find applications in various fields:

• Image and speech recognition
• Natural language processing
• Autonomous vehicles
• Medical diagnosis
• Financial forecasting
• Game playing and strategy

10. Challenges and Future Directions:

Some ongoing challenges and areas of research include:

• Improving energy efficiency and reducing computational requirements (Sarwar et al., 2016, pp. 145–150)
• Enhancing interpretability and explainability of neural network decisions
• Developing more biologically plausible neuron models
• Exploring novel architectures and learning algorithms for specific tasks

In conclusion, artificial neurons are the fundamental units of artificial neural networks, inspired by biological neurons but simplified for computational efficiency. They continue to evolve, with researchers exploring new designs, implementations, and applications to push the boundaries of artificial intelligence and neuromorphic computing.

Artificial Neuron in Layman Term

1. Receiving information: The artificial neuron receives pieces of information, just like how you might receive messages from your friends. These pieces of information are called “inputs“ (Explain like I’m Five: Artificial Neurons | by Maurice Henry Buettgenbach | Towards Data Science, n.d.).
2. Weighing the importance: Not all information is equally important, right? So, the artificial neuron gives different levels of importance to each piece of information it receives. This is like how you might pay more attention to a message from your best friend than a random advertisement (Explain like I’m Five: Artificial Neurons | by Maurice Henry Buettgenbach | Towards Data Science, n.d.).
3. Adding it all up: The neuron then adds up all this weighted information. It's like collecting all the important messages you've received throughout the day (Explain like I’m Five: Artificial Neurons | by Maurice Henry Buettgenbach | Towards Data Science, n.d.).
4. Making a decision: Now comes the crucial part. The neuron has to decide whether this combined information is important enough to pass along. It's like you deciding whether the news you've heard is exciting enough to tell your other friends (Explain like I’m Five: Artificial Neurons | by Maurice Henry Buettgenbach | Towards Data Science, n.d.).
5. Passing it on (or not): If the combined information reaches a certain level of importance (called a “threshold“), the neuron gets “excited“ and passes the information along to other neurons. If not, it stays quiet (Explain like I’m Five: Artificial Neurons | by Maurice Henry Buettgenbach | Towards Data Science, n.d.).

This whole process happens incredibly fast and many, many times across a network of these artificial neurons. It's this network of neurons passing information around and making decisions that allows artificial intelligence to learn and make complex decisions, much like how the neurons in our brains work together to help us think and learn (Explain like I’m Five: Artificial Neurons | by Maurice Henry Buettgenbach | Towards Data Science, n.d.; What Is an Artificial Neuron? | Definition from TechTarget, n.d.).

The beauty of artificial neurons is that they can be trained to recognize patterns and make decisions about all sorts of things – from identifying objects in images to predicting weather patterns or even playing chess (What Is an Artificial Neuron? | Definition from TechTarget, n.d.).

So, in essence, an artificial neuron is like a tiny decision-maker in a vast network, working together with countless others to solve complex problems and learn from experience, just like the neurons in our own brains!

There might be more research done in the future. Stay tuned.

You might be interested in What is Artificial Intelligence, What Are The Top 10 Emerging Technologies – AI 1 of Them

References

V, Y. (2024). From McCulloch to GPT – 4: stages of development of artificial intelligence. Artificial Intelligence.

Elamin, M. O. I. (2024). AI Through the Ages: Unlocking Key Opportunities and Navigating Challenges in the History and Future of Artificial Intelligence. International Journal of Religion.

Al-Amin, Md., Ali, M. S., Salam, A., Khan, A., Ali, A., Ullah, A., Alam, M. N., & Chowdhury, S. K. (2024). History of generative Artificial Intelligence (AI) chatbots: past, present, and future development. arXiv.Org, abs/2402.05122.

Sharma, R. (2020). Artificial Intelligence in Healthcare: A Review. Turkish Journal of Computer and Mathematics Education.

Kufel, J., Bargieł-Łączek, K., Kocot, S., Koźlik, M., Bartnikowska, W., Janik, M., Czogalik, Ł., Dudek, P., Magiera, M., Lis, A., Paszkiewicz, I., Nawrat, Z., Cebula, M. M., & Gruszczyńska, K. (2023). What Is Machine Learning, Artificial Neural Networks and Deep Learning?—Examples of Practical Applications in Medicine. Diagnostics, 13.

Rathore, P., Dadich, N., Jha, A., & Pradhan, D. (2018). Effect of Learning Rate on Neural Network and Convolutional Neural Network. International Journal of Engineering Research and Technology.

Beniaguev, D., Segev, I., & London, M. (2019). Single cortical neurons as deep artificial neural networks. Neuron, 109, 2727-2739.e3.

Katayama, H., Fujii, T., & Hatakenaka, N. (2018). Theoretical basis of SQUID-based artificial neurons. Journal of Applied Physics.

Yan, X., Shao, Y., Fang, Z., Han, X., Zhang, Z., Niu, J., Sun, J., Zhang, Y., Wang, L., Jia, X., Zhao, Z., & Zhen-Guo. (2023). A low-power reconfigurable memristor for artificial neurons and synapses. Applied Physics Letters.

Lu, Y., Li, Y., Li, H., Wan, T.-Q., Huang, X., He, Y., & Miao, X. (2020). Low-Power Artificial Neurons Based on Ag/TiN/HfAlOx/Pt Threshold Switching Memristor for Neuromorphic Computing. IEEE Electron Device Letters, 41, 1245–1248.

Yang, S., Shin, J., Kim, T., Moon, K., Kim, J., Jang, G., Hyeon, D., Yang, J., Hwang, C., Jeong, Y., & Hong, J. (2021). Integrated neuromorphic computing networks by artificial spin synapses and spin neurons. NPG Asia Materials, 13, 1–10.

Kotsovsky, V., Geche, F., & Batyuk, A. (2015). Artificial complex neurons with half-plane-like and angle-like activation function. International Conference on Computer Science and Information Technologies, 57–59.

Noel, M. M., Bharadwaj, S., Muthiah-Nakarajan, V., Dutta, P., & Amali, G. B. (2021). Biologically Inspired Oscillating Activation Functions Can Bridge the Performance Gap between Biological and Artificial Neurons.

Sarwar, S. S., Venkataramani, S., Raghunathan, A., & Roy, K. (2016). Multiplier-less Artificial Neurons exploiting error resiliency for energy-efficient neural computing. Design, Automation and Test in Europe, 145–150.

Explain like I’m five: Artificial neurons | by Maurice Henry Buettgenbach | Towards Data Science. (n.d.). Retrieved November 16, 2024, from https://towardsdatascience.com/explain-like-im-five-artificial-neurons-b7c475b56189

What is an artificial neuron? | Definition from TechTarget. (n.d.). Retrieved November 16, 2024, from https://www.techtarget.com/searchcio/definition/artificial-neuron