For the past year, I have been working on a technology company focused in the AI (artificial intelligence) space. The only thing constant about AI is its rapid pace of change. Chat GPT, which was most consumers’ introduction to AI, just turned one year old, and since then, the feverish pace at which AI has moved, board drama and all, has been incredible.
But what about the basics? What should you know as a business person, or a casual follower of AI to catch you up on what you need to know? I’ve compiled a primer aimed at guiding you back to the foundational elements of AI, so you can be conversant and knowledgeable on the fundamentals. In future parts of this series, I’ll be addressing practical use cases of AI and companies you should follow.
A previous Forbes contributor put a similar guide together 6 years ago. However, considering the dramatic advancements since then, an updated version seemed not just appropriate, but necessary.
Glossary Of Terms
I wanted to first start off with a key set of terms that you should know in order to understand what artificial intelligence is, including AI itself. Of course many define these terms differently, so I recommend you do your own research and deeper dives on the subject. But to start with, I recommend you know the following:
- Artificial Intelligence (AI): The broad concept of machines being able to carry out tasks in a way that we would consider smart or human-like.
- Machine Learning: A subset of AI focused on algorithms that can learn from data, identify patterns, and make decisions with minimal human intervention. You might hear terms like “reinforcement learning” here where humans actually make algorithms better.
- Deep Learning: A specialized machine learning technique based on artificial neural networks with multiple layers that can process enormous datasets and power innovations like computer vision and natural language processing. Frankly speaking, this area of AI is extremely technical and way above my head.
- Neural Networks: Computing systems, modeled after the neural networks in the human nervous system, that can progressively learn and become more accurate at tasks like object recognition in images.
- Natural Language Processing (NLP): The branch of AI focused on enabling computer systems to understand, interpret, communicate, and manipulate human languages. For example, if you are a power user of Chat GPT, you are exposed to this on a regular basis.
- Computer Vision: The field of AI focused on enabling computers and systems to identify, analyze, classify, and understand digital images and videos. For those that use Google Image search, this is a very practical use case for computer vision.
- Artificial General Intelligence (AGI): Hypothetical AI that demonstrates human-level intelligence and capabilities across a variety of cognitive tasks. This is a hot-button subject in the world of AI–will machines eventually become smarter than us? There is not a clear definition of when AGI is “achieved.”
What Are The Core Technologies Powering AI?
So you want to start an AI company, and don’t know the different areas that need support? Or you are seeking investment ideas for potential companies to target? There are different areas of AI that present opportunities for investment, pathways for employment, or avenues for further education and understanding.
- Data: You cannot have AI without vast amounts of data. It’s a precursor, and a hot topic subject, given Sarah Silverman’s recent Lawsuit against Open AI and Meta for allegedly using copyrighted material for their models. Companies leveraging vast datasets: Meta, Google, Microsoft, Amazon, Tesla, Waymo
- Computing Power: Of course Nvidia is the darling here as of late, but you need a lot of computing power to process data. Due to chip shortages, companies are scrambling to secure enough computing resources to process data. Chipmakers providing hardware for model training: Nvidia, Intel, AMD, Qualcomm Cloud infrastructure companies: AWS, Microsoft Azure, Google Cloud
- Algorithms: Advanced algorithms, like those developed by DeepMind, Open AI and Anthropic, provide the core logic and step-by-step calculations enabling AI systems to learn. For example, various types of machine learning algorithms allow AI models to enhance their performance in tasks through exposure to large sets of quality data. Algorithm innovators: DeepMind, Anthropic, OpenAI, Cohere
- Modeling: Engineers test and refine machine learning models, like neural networks, to accurately power capabilities such as computer vision and natural language understanding. For instance, the task of correctly labeling images. Leading modeling frameworks: TensorFlow (Google), PyTorch (Meta), MXNet (Amazon)
- Application Interface: This is where you probably have the most exposure. Apps like Chat GPT allow you to interact with GPT-4 and produce results. Consumer tech leaders building AI applications: Meta, Apple, Amazon, Microsoft, Open AI, Anthropic, MidJourney, Runway ML
Finally, large language models (LLMs) like GPT-4 have emerged as a key component powering conversational AI. LLMs provide the underlying language capabilities leveraged by chatbots and voice assistants.
If you really want to get into the guts of how AI functions, consider exploring these detailed technical papers. These selections not only serve as excellent entry points but are also widely recognized as pivotal contributions to the field. However, bear in mind, this list is not exhaustive:
For a frequently updated, high-level perspective on AI, I recommend Rowan Cheung’s newsletter The Rundown.
I will be adding to this series frequently, so please stay tuned.
Read the full article here