AI Term of the Moment
high bandwidth flash
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 large AI networks attempt to 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 the early voice assistants in smartphones, AI also provides the extraordinary service of generating content of every variety either from scratch or following uploaded samples (see generative AI).
Not Ordinary Programming
Nothing like ordinary application programming, modern AI uses a "neural network" made up of layers that connect to each other mathematically. 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 "inference," must also be designed and fine-tuned. The inference part 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 data center time, power and electricity. See AI data center.
A Simple Pattern Recognition Example
If 10,000 chest x-rays showing lung cancer and 10,000 cancer-free x-rays are fed into a neural network, such systems can often detect the disease better than medical professionals, and most importantly, faster. A 2025 article in the Washington Post stated that two-thirds of all radiology departments in the U.S. are using AI to augment their decision making. The Mayo Clinic claimed its AI model can help detect pancreatic cancer up to three years earlier. Medical care may be "the" best breakthrough from AI. 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 Author Alan Freedman
I've been in the information technology industry more than 65 years and have written about every important computer product and concept. I have also defined more than a thousand AI terms. Therefore, I have a right to an opinion about AI, and it is simply this: "everybody is right!" AI is likely to make monumental 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 the early voice assistants in smartphones, AI also provides the extraordinary service of generating content of every variety either from scratch or following uploaded samples (see generative AI).
Not Ordinary Programming
Nothing like ordinary application programming, modern AI uses a "neural network" made up of layers that connect to each other mathematically. 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 "inference," must also be designed and fine-tuned. The inference part 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 data center time, power and electricity. See AI data center.
A Simple Pattern Recognition Example
If 10,000 chest x-rays showing lung cancer and 10,000 cancer-free x-rays are fed into a neural network, such systems can often detect the disease better than medical professionals, and most importantly, faster. A 2025 article in the Washington Post stated that two-thirds of all radiology departments in the U.S. are using AI to augment their decision making. The Mayo Clinic claimed its AI model can help detect pancreatic cancer up to three years earlier. Medical care may be "the" best breakthrough from AI. 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 more than 65 years and have written about every important computer product and concept. I have also defined more than a thousand AI terms. Therefore, I have a right to an opinion about AI, and it is simply this: "everybody is right!" AI is likely to make monumental changes for good and for bad. See Asilomar Conference on Beneficial AI.