Four Flavors of AI Through the Eyes of ChatGPT

"Early adoption of these innovations can drive significant competitive advantage and business value and ease problems associated with the fragility of AI models."

The ChatGPT phenom is historic for those who generate content - this is certain.  But it doesn't take too much to see the linguistics app is just the tip of the iceberg.  It's more than asking Siri for directions and receiving unsolicited weather updates from Alexa.

• Tell it what you want to do on a spreadsheet and it will write the macro.
• Tell it what you want to program for the interwebs and it will write the HTML.
• Ask it to analyze a 1000-word article about the four types of AI, summarize it into a 500-word blog, and write both a Tweet and a LinkedIn post intro and it does.

Ask it to apply the previous content and write an article on the impact of AI on the Office Technology industry - in the voice of Hemingway - and then put together an outline, and agenda, for a show, with sample questions, about How a copier salesperson can use ChatGPT to find more prospects and sell more copiers and document management software solutions?

I did and it did.

_____

I pulled this off of the Gartner site:

"The 2022 Gartner Hype Cycle™ for Artificial Intelligence (AI) identifies must-know innovations in AI technology and techniques that go beyond the everyday AI already being used to add intelligence to previously static business applications, devices, and productivity tools.

“Notably, the AI Hype Cycle is full of innovations expected to drive high or even transformational benefits,” says Afraz Jaffri, Director Analyst at Gartner. “Pay particular attention to innovations expected to hit mainstream adoption in two to five years, including composite AI, decision intelligence, and edge AI. Early adoption of these innovations can drive significant competitive advantage and business value and ease problems associated with the fragility of AI models.”

The AI innovations on the Hype Cycle reflect complementary and sometimes conflicting priorities across four main categories:

"𝟭. 𝗗𝗮𝘁𝗮-𝗰𝗲𝗻𝘁𝗿𝗶𝗰 𝗔𝗜

A focus on enhancing and enriching the data used to train the algorithm as opposed to tweaking the AI models themselves.

𝗦𝘆𝗻𝘁𝗵𝗲𝘁𝗶𝗰 𝗱𝗮𝘁𝗮 is a great example of data-centric AI. It avoids using personally identifiable information, reduces cost and saves time in ML development, and improves ML performance.

𝟮. 𝗠𝗼𝗱𝗲𝗹-𝗰𝗲𝗻𝘁𝗿𝗶𝗰 𝗔𝗜

Despite the shift to data-centric AI, AI models still need attention. Innovations here include physics-informed AI, composite AI, generative AI, foundation models, and deep learning.

An example of 𝗽𝗵𝘆𝘀𝗶𝗰𝘀-𝗶𝗻𝗳𝗼𝗿𝗺𝗲𝗱 𝗔𝗜 is using First-Principles Simulations together with reinforcement learning to unlock thousands of use cases the significantly narrow the gap between the digital and physical worlds.

𝟯. 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀-𝗰𝗲𝗻𝘁𝗿𝗶𝗰 𝗔𝗜

Innovations here include AI engineering, decision intelligence, operational AI systems, ModelOps, AI cloud services, smart robots, natural language processing (#NLP), autonomous vehicles, intelligent applications, and computer vision.

𝗘𝗱𝗴𝗲 𝗔𝗜, for example, refers to the use of AI techniques embedded in Internet of Things (IoT) endpoints, gateways and edge services, in applications ranging from autonomous systems to streaming analytics.

Pay particular attention to innovations expected to hit mainstream adoption in two to five years, including composite AI, decision intelligence, and edge AI. 

𝟰. 𝗛𝘂𝗺𝗮𝗻-𝗰𝗲𝗻𝘁𝗿𝗶𝗰 𝗔𝗜

When AI replaces human decisions, it amplifies good and bad outcomes alike. Responsible AI enables the right outcomes by resolving dilemmas rooted in delivering value versus tolerating risks (i.e. Reid Blackman,'s AI for not bad).

Other innovations in this group include AI trust, risk and security management (TRiSM), digital ethics, and AI maker and teaching kits."

Digested and summarized by SkyNet:

Data-centric AI is when the focus is on making the data used to train the algorithm better, instead of just changing the AI itself. An example is synthetic data which helps avoid using personal information and makes the process of creating AI cheaper and faster.

Model-centric AI is when the focus is on improving the AI models themselves. This can include using things like physics and deep learning to make the AI more accurate.

Application-centric AI is when the focus is on how the AI is used in real-world situations. This can include things like using AI in self-driving cars or natural language processing.

Human-centric AI is when the focus is on making sure that the AI is used in a responsible way so that it does not cause any harm. This can include things like making sure the AI is fair and ethical and managing any risks that might come with using AI.

The Four Categories of AI. As told by SkyNet, in the voice of Thomas Edison:

"Ah, my dear friend, the field of AI is ever-evolving. We must keep our eyes peeled for ways to improve our data, models, applications, and the human impact of our creations.

In the realm of data-centric AI, we must strive to enhance the quality of the information used to instruct our machines, just as one sharpens a tool to increase its precision.

In the realm of model-centric AI, we must explore new frontiers such as physics-informed AI and deep learning, improve the performance of our machines.

In the realm of application-centric AI, we must be creative in applying our inventions to the world around us, from self-driving cars to natural language processing.

And in the realm of human-centric AI, we must ensure that our machines will be responsible servants and not masters, always striving for the right outcomes and managing any potential risks."

_______

I like T. Edison's rendition.