{"id":30941,"date":"2025-09-26T22:59:44","date_gmt":"2025-09-26T22:59:44","guid":{"rendered":"https:\/\/drdavidcring.net\/descartes-ideas\/?p=30941"},"modified":"2025-12-09T21:55:13","modified_gmt":"2025-12-09T21:55:13","slug":"how-to-use-ai-models","status":"publish","type":"post","link":"https:\/\/drdavidcring.net\/descartes-ideas\/ai\/how-to-use-ai-models\/","title":{"rendered":"Using AI in Humanities Research"},"content":{"rendered":"\n
<\/div>\n\n\n\n

Sponsored by the NAU Institute for Advancing Applications in Artificial Intelligence (IAAAI)<\/mark><\/h2>\n\n\n\n

Wednesday, October 15, 2025, 4\u20135 pm in HRM 136 on the NAU campus in Flagstaff, AZ.<\/mark><\/h2>\n\n\n\n
<\/div>\n\n\n\n

\"An<\/p>\n\n\n\n

<\/div>\n\n\n\n

\/ai\/how-to-use-ai-models<\/a><\/u><\/span><\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

\"A<\/p>\n\n\n\n

<\/div>\n\n\n\n

THE CHOICE<\/mark><\/h2>\n\n\n\n
<\/div>\n\n\n\n

Imagine, if you will, that there are two doors, and you must choose and enter one or the other. Each door has a name. Door One<\/strong> is called \u201cDON\u2019T WANT TO<\/strong>,\u201d and the other door is called \u201cLET\u2019S FIND OUT<\/strong>.\u201d<\/span><\/p>\n\n\n\n

If you choose Door One \u201cDON\u2019T WANT TO\u201d<\/strong> when you step through it, everything seems just as it was before you stepped through.<\/span>

\u201cWhew, you say, what a relief!\u201d<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

But maybe NOT!<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

\"The<\/p>\n\n\n\n

<\/div>\n\n\n\n

However, before you step through \u201cDON\u2019T WANT TO\u201d<\/strong> or \u201cLET\u2019S FIND OUT,\u201d<\/strong> you need to know what\u2019s behind door number two. Behind this door, you will discover a box with a genie<\/a> (or Jinn<\/a>) inside, named:<\/span>

HAL<\/a><\/strong> (spelled in all capital letters\u2014we don\u2019t know why<\/em><\/strong>).<\/span><\/p>\n\n\n\n

[Hint:<\/strong> HAL<\/a> (H<\/strong>euristically<\/a> Programmed Al<\/strong>gorithmic<\/a> Computer)]<\/p>\n\n\n\n

<\/div>\n\n\n\n

https:\/\/share.icloud.com\/photos\/055Lp6l8nv59TK5LCTl99vy9g<\/a><\/p>\n\n\n\n

\u201cOpen the Podbay Doors<\/u><\/span><\/a>\u201d starring Dr. David C. Ring<\/u><\/span><\/a> \"A<\/a>and thanks to the voice of Keir Dullea<\/u><\/span><\/a> \"An<\/a><\/h1>\n\n\n\n
<\/div>\n\n\n\n

Anyway, HAL<\/a> the Jinn\/genie<\/a>, known as being neither good nor evil (according to Wikipedia<\/a>!), and who willingly offers you not a measly three wishes, but for the rest of your life, you can converse with it using any and all questions or prompts and it will tell you the answer. Forever and unlimited. Although the free versions are just demo\u2019s when the tools you should be using require payment.<\/span>

\u201cMostly, no free lunches,\u201d HAL<\/a> informs us.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Sooooo, the question before you is which door ought you to choose?<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

I say ought intentionally because it immediately raises ethical concerns. David Hume the eighteenth-century Scottish philosopher made famous the \u201cis\/ought\u201d distinction claiming that one cannot derive an ought from what is so that descriptive statements never logically entail anything whatsoever about prescriptive ones. For example, simply because humans evolved to eat meat (a biological fact) does not entail that humans “ought” to eat meat (a moral value).<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

My intuitive sense would be that humans are an inquisitive lot. We like to learn and know new things.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

AI Joke:<\/strong>  \u201cHumans are inquisitive? Absolutely\u2014we invented curiosity killed the cat, but AI revived it with nine lives of generated content. In humanities, that\u2019s called \u2018iterative research\u2019!\u201d<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

[Parenthetical Aside:<\/strong> I had never heard of iterative research so I Googled it.]<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

\"A<\/a><\/p>\n\n\n\n

<\/div>\n\n\n\n

If HAL<\/a> can answer any questions about anything, it might seem stupid not to ask him some questions.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

\u201cWhy not, you say?\u201d<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Well, probably asking him some questions cannot be that bad in terms of consequences, but what about consequences from acting on the information HAL provides you?<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n
\n\n\n\n
<\/div>\n\n\n\n

BAD CONSEQUENCES FROM USING AI<\/mark><\/h2>\n\n\n\n
<\/div>\n\n\n\n

DIGRESSION:<\/mark><\/strong> What have been some bad consequences from using AI. We’re any of them of any consequence?<\/h2>\n\n\n\n
<\/div>\n\n\n\n

\"A<\/p>\n\n\n\n

<\/div>\n\n\n\n

\u201d32 times artificial intelligence got it catastrophically wrong<\/strong>\u201d<\/a><\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Saving the Google ALPHABET story<\/strong> for last, consider these five problems caused from using AI after stepping through our Door Two:<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Boeing 737 MAX MCAS System (2018-2019)<\/mark><\/h2>\n\n\n\n

The AI-augmented flight control software contributed to two fatal crashes by overriding pilot inputs based on faulty sensor data, killing 346 people<\/strong> and grounding the fleet worldwide for nearly two years.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Tesla Autopilot Crashes (Multiple, 2016-2025)<\/mark><\/h2>\n\n\n\n

Tesla’s semi-autonomous driving system has been involved in fatal accidents where it failed to detect obstacles like emergency vehicles or pedestrians, resulting in recalls, lawsuits, and regulatory probes<\/strong> into overreliance on AI.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

IBM Watson for Oncology (2010s-2020s)<\/mark><\/h2>\n\n\n\n

Marketed as a revolutionary cancer treatment advisor, Watson provided unsafe or incorrect recommendations in many cases, leading to project cancellations by hospitals and IBM’s eventual discontinuation after billions in investment<\/strong>.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

GM Cruise Robotaxi Incident (2023)<\/mark><\/h2>\n\n\n\n

A self-driving Cruise vehicle struck and dragged a pedestrian<\/strong> after failing to properly detect her, causing the suspension of operations in San Francisco, permit revocations, and federal investigations<\/strong>.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Zillow’s iBuying Algorithm (2021)<\/mark><\/h2>\n\n\n\n

Zillow’s AI-driven house-flipping tool overpaid for properties based on inaccurate market predictions, causing massive financial losses and the layoff of 2,000 employees <\/strong>when the program was shut down.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

Google Bard’s James Webb Space Telescope Error (2023)<\/mark><\/h2>\n\n\n\n

Google’s AI chatbot Bard<\/strong> incorrectly stated in a promotional demo that the James Webb Space Telescope took the first pictures of an exoplanet outside our solar system<\/strong>, leading to a roughly 8% drop in Alphabet’s stock<\/strong> and<\/span> a $100 billion loss in market value.<\/strong><\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n
\n\n\n\n
<\/div>\n\n\n\n

ASKING QUESTIONS<\/mark><\/h2>\n\n\n\n
<\/div>\n\n\n\n

We probably should start out by asking HAL how accurate and correct his answers are.<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

US:<\/strong> How accurate and correct are your answers to queries or prompts?<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n

ChatGPT 5:<\/strong> The accuracy and correctness of my answers vary depending on several key factors:<\/mark><\/h2>\n\n\n\n
<\/div>\n\n\n\n

I. <\/strong>General Accuracy Level<\/mark><\/h2>\n\n\n\n
<\/div>\n\n\n\n

Empirical studies and OpenAI\u2019s evaluations suggest that I (GPT-4o and GPT-4-turbo, October 2023 and later variants) achieve very high accuracy in domains such as:<\/span><\/p>\n\n\n\n

<\/div>\n\n\n\n