Stream of Consciousness on AI, Technostress, and Faith

🌱Seed 🙂Agree

---

Importance: 10%

The Bid Idea

Collection of random Thoughts about AI, Faith and Techno Stress

---

Tip

Start on more structured ideas: A Theology of Technology Principles for Working with Tech Tools

Technostress is not being able to cope with the pace of change

Digital overload not being able to cope with the amount of data and communication getting thrown at you 

Technophobia physical and mental stress due to being confronted with new technology at a rate faster than you can manage or keep up with. 

Advent of LLM and ChatGPT, LangChain, Stable-Diffusion

How to we cope with change in a healthy way

How to we keep faith in the center in the face of techno salvation and techno end times

What is the proper place for technology 

Role of human memory in the face of machine memory  

The role of technology often comes down to the simple discussion of what kind of work can be offloaded from humans and entrusted to the technology. The ethical question then becomes SHOULD that work be offloaded? What are the benefits of offloading it? What do humans lose by not being directly responsible for this part of the work?

Another line of inquiry is to think about the growing reality that we live in a world where a particular person needs to keep in their minds less and less information in order to be functional. How does living in a world where I don’t need to know much to survive affect the way we live and interact with each other and the world? 

I think that this helps us realize the importance of memory and learning Scripture. In a world where we can offload so much information and work to technology the truly important things are the things of faith and life flowing from that.

The Preservation of Tail Knowledge

The conversation that government and society affects or lives is one that is not unfamiliar. Laws and social movements put pressure on us and our loved ones to think and live a certain way. Technology is connected with this in part but I think we often fail to see the underlying narratives that drive the adoptions and development of many technologies. 

Open source philosophy 

Is AI making your life easier or just giving you one more thing to do?

Why AI? Why now?

It’s not going away

We have been using it for a long time without even thinking about it

Google Translate, etc.

General AI Best Practices

Check for bias and accuracy: always double check before sharing with students.

The 80-20 approach: Use AI for initial work, but make sure to add your final touch, review for bias and accuracy, and contextualize for the last 20.

Production of rubrics

Production as assignments

Time management for teachers

The last 20% is vital for teachers - they have to be the ones to complete it. The 80% teaches, the 20% puts it into practice.

Our judgment matters

Know the limit - things may be out-of-date. It may represent a worldview that we don’t want to tell our students something we don’t agree with. Know the ultimate source of knowledge.

Protect Privacy - never use personal information or student personal information.

Teacher Workflow

Chat GPT (free)

The prompt you put in dictates how well of a response you will get back.

Plug in an idea for a lesson and add CC standards to the program

Magic School Lesson Plan (free, but there is a pro plan at $100 per year)

Takes an idea and makes an entire lesson plan

Made for teachers

Creates assignments

Following the language model teachers have been trained in (objectives, assessments, opening, intro, guided practice, independent practice, etc.)

Gives opportunity to ask questions about the lesson

Support for differentiated learning needs through text leveler - you can even get texts from outside sources, copy and paste, and make it at a different reading level.

Even can use Lexile levels in some situations.

Creates different types of questions that can be used at all levels.

Give teachers the platform for creating rubrics based on the lesson that was just developed.

Gamma Presentation (free)

Produces documents, presentations, and websites

Good platform to begin a presentation with options for tailoring it to you.

Creates an outline before the presentation is developed.

Curipod Interactive (free)

Interactive slides like Peardeck

Used AI to look at the presentation you already have and make it interactive (curify my slides)

Magic School Investigation

Inservice worthy?

This is a good place to start with teachers because it is:

Applicable PK-8

Simple to use

Free

Useful for missing topics or lessons within an established curriculum

Look into:

How can this be used in reading?

Theology strong?

Other topics

AI Ethics

Academic Integrity

Student use of AI

“AI Proofing” Assignments

Transfer of Learning Issues

Prompt based use of LLM and image generation like fabric

Agent based use like AGiXT

Data based use like quiver

Basic two perspective humans will be replaced by AGI

Humans will be augmented by useful ai tools and assistants

Specific fine tuned models that out perform GPT-4 in specific tasks https://huggingface.co/predibase

-   Video on how digital computers are hitting their limits - Why the Future of AI & Computers will be Analog

-  Analog can compute anything that can be expressed as a differential equation

-  A differential equation calculates describe how something changes

Using AI Well

From Unsupervised learning:

1. I prefer using APIs for any programmatic workflows. 2. For day-to-day use I recommend Fabric because it’s super flexible, has tons of use-cases (Patterns), and lets you use OpenAI, Anthropic, and local models via Ollama. 3. For super quick demos of open models, plus some API functionality, I like ollama.com. 4. For using newer, cutting-edge models, my favorite tool is LMStudio.
5. The big advantage of LMStudio is that you can use the latest models. Including for example, uncensored local models, like my current favorite—QuantFactory/Meta-Llama-3-70B-Instruct-GGUF 6.  You just type in the model you want, and you don’t even need to go get it from Huggingface. It downloads it for you right in the tool, and you can use it right there in the tool, because it’s also a prompting interface.

From Techoasis:

Chatting with Your Data
As the name implies, Generative AI models are ‘generative.’ But what does that mean? Well, simply put, they are trained to ‘regenerate’ the training data.
In other words, their success depends on how well they have learned the training data and can regenerate it. In layman’s terms, they are trained to imitate the training data.
This is why frontier AI models aren’t that intelligent today. Just like memorizing math theorems doesn’t make you a mathematician, imitating human language doesn’t make you a human-level ‘being.’\
Therefore, the most important principle when using GenAI products is to use them only in situations requiring knowledge elicitation.
But what do I mean by that?
One way you are guaranteed to have a great experience is in ‘Chat with your data’ use cases. In other words, you hand the model data, via PDF or prompt, and you question the model about it.
Today’s models excel at in-context learning, meaning they can efficiently work with data they haven’t seen before. However, don’t get too carried away with this capability; use it wisely.
In layman’s terms, ask questions about the data; don’t use that data to ask questions beyond it.
In other words, the fact that the model can process the data in your PDF file doesn’t mean that it can use that data to generate new thoughts or insights.
Remember, it’s all imitation, and nothing is new or ‘emergent,’ as some in Silicon Valley will want you to believe.
The bottom line is that any use cases focusing on knowledge management and elicitation are a great suitor for GenAI today.
Leverage the Training
As models are trained on imitation, research labs have modified their training datasets so that the model imitates certain behaviors. This is absolutely key.
Nevertheless, even though they are considered general-purpose models and thus can generalize into multiple tasks… they can as long as they have been well-trained in them.
Luckily, just taking a quick look at some of the open datasets used to train these models will give you great insights into how to use them: