Term of the Moment
broadcast storm
Definition: AI in a nutshell
AI is essentially pattern recognition, which identifies objects and natural language, as well as the trends in every discipline on the planet. The largest AI systems attempt to recognize and analyze the world's knowledge.
What AI Does
In medicine and science, AI is used to explore the interactions in every molecular structure. AI is used to make marketing and financial forecasts and answer questions about almost everything. A huge step beyond voice assistants in every phone, AI also provides the extraordinary service of generating content of every variety starting from scratch or following uploaded samples (see generative AI).
Not Ordinary Programming
Nothing at all like ordinary application programming, most AI applications use a "neural network" made up of multiple layers that connect to each other mathematically, which loosely mimics the human nervous system (neural system). The neural network is designed and programmed by AI engineers to be a "model" that is trained and fine-tuned on huge amounts of data. The execution part that users deal with, known as the "inference engine," must also be designed and fine-tuned. The inference engine employs the model to answer questions and generate content for the user. See AI programming, AI training vs. inference and AI secret sauce.
Models Are Trained on the World's Information
The data samples used for training come from websites, blogs, articles, dictionaries, encyclopedias and books, essentially all the information ever published online. The training phases can take a huge amount of datacenter time, power and electricity.
A Simple Example of Pattern Recognition
If 10,000 chest x-rays showing lung cancer and 10,000 cancer-free x-rays are fed into a neural network, the system learns the differences between them. Such systems can detect diseases better than medical professionals, and most importantly, faster. An article in the Washington Post in 2025 stated that two-thirds of all radiology departments in the U.S. are using AI. The Mayo Clinic detects pancreatic cancers up to three years before a clinician can. See large language model, deep learning and neural network.
Concern for the Future
What worries people is the research being done to replace human decision making with AI. For example, should AI be used to expand a company or pull back? Even more significant, do we leave the decision to go to war up to a machine? There is a huge amount of controversy regarding AI and the future (see AI anxiety). See AI, technology singularity, AI hallucination, AGI and AI stages.
AI Will Reign Supreme
A Note from the Author
I've been in the information technology industry for more than 65 years and have added over a thousand AI terms to this reference. Therefore, I have a bona fide right to an opinion, and my opinion is that "everybody is right!" AI is likely to make catastrophic changes for good and for bad. See Asilomar Conference on Beneficial AI.
What AI Does
In medicine and science, AI is used to explore the interactions in every molecular structure. AI is used to make marketing and financial forecasts and answer questions about almost everything. A huge step beyond voice assistants in every phone, AI also provides the extraordinary service of generating content of every variety starting from scratch or following uploaded samples (see generative AI).
Not Ordinary Programming
Nothing at all like ordinary application programming, most AI applications use a "neural network" made up of multiple layers that connect to each other mathematically, which loosely mimics the human nervous system (neural system). The neural network is designed and programmed by AI engineers to be a "model" that is trained and fine-tuned on huge amounts of data. The execution part that users deal with, known as the "inference engine," must also be designed and fine-tuned. The inference engine employs the model to answer questions and generate content for the user. See AI programming, AI training vs. inference and AI secret sauce.
Models Are Trained on the World's Information
The data samples used for training come from websites, blogs, articles, dictionaries, encyclopedias and books, essentially all the information ever published online. The training phases can take a huge amount of datacenter time, power and electricity.
A Simple Example of Pattern Recognition
If 10,000 chest x-rays showing lung cancer and 10,000 cancer-free x-rays are fed into a neural network, the system learns the differences between them. Such systems can detect diseases better than medical professionals, and most importantly, faster. An article in the Washington Post in 2025 stated that two-thirds of all radiology departments in the U.S. are using AI. The Mayo Clinic detects pancreatic cancers up to three years before a clinician can. See large language model, deep learning and neural network.
Concern for the Future
What worries people is the research being done to replace human decision making with AI. For example, should AI be used to expand a company or pull back? Even more significant, do we leave the decision to go to war up to a machine? There is a huge amount of controversy regarding AI and the future (see AI anxiety). See AI, technology singularity, AI hallucination, AGI and AI stages.
MIT professor Max Tegmark's best-selling book postulates an AI that far exceeds human intelligence and literally takes over.
I've been in the information technology industry for more than 65 years and have added over a thousand AI terms to this reference. Therefore, I have a bona fide right to an opinion, and my opinion is that "everybody is right!" AI is likely to make catastrophic changes for good and for bad. See Asilomar Conference on Beneficial AI.