Another Think Coming

Image source: Someone in a Facebook ChatGPT for Teachers group

Great news, right? Custom GPTs will be usable by free users of ChatGPT, those custom GPTs will no longer be behind a paywall. The letter above was shared by an educator who is creating custom GPTs for his K-12 students to use.

Can vs Should

Just because you can use a technology in a K-12 classroom, should you? In a blog entry that goes to the hearts of the values mismatch between the mission of schools and AI development fraught with a morass of ethical issues, Tom Mullaney explores the issues.

Is AI Compatible with Schools?

That question is the title of Tom’s blog entry. In it, he asks the question many have been asking as they watch AI tech companies make a run at school funding, a mass charge to bring AI into schools, and charge every person who uses it, at the same time using that data for AI model training.

His blog entry highlights several issues with AI such as equity, racism, bias, copyright, human exploitation, AI lies or hallucinations. To this pile, I would add climate change impact, water usage, destruction of the environment to build massive data centers.

AI Bits: What Are We Worried About?

Tom isn’t the only one with concerns. John Dolman makes a list of concerns, but doesn’t go into details, by his own admission. His list includes:

• Plagiarism and academic dishonesty. This is on target. Someone tells you not to use AI, you do, and now you have academic dishonesty. It pales in significance to the next bullet, in my opinion [Miguel]
• Skill reduction. This is that ubiquitous, and valid, concern that AI supplants student thinking, resulting in students who can’t think their way out of a paper bag, put their thinking down in writing, or use writing to clarify their thinking.
• Misinformation. This is the AI lies, hallucination, factual inaccuracy.
• Intellectual Property and Data Privacy. This is copyright, but also dropping all the privacy data from students, staff, schools into AI could result in information popping up somewhere else and you never know about it.
• Equity and access. We’re seeing this already, the fact that Teacher A has access to super-powered, customized AI (e.g. MagicSchool, Diffit, Brisk, Eduaide,Teachaid, etc.) while someone else is using lame, er, Llama, via Meta or struggling along with a free AI. The quality of the responses is markedly different, better on these other solutions.
• Pedagogical impact. What IS the impact on pedagogy and education? Some see AI as the tool to dismantle public education (stuck with bottom tier AIs) while private, charter schools (with deep pockets funded AI tools available to the rich) get the benefit of anything goes AI. It’s a “new arms race” for education, and AI is already being “weaponized” to give advantage to the privileged elite, whomever they might be, over the regular working folks and their kids forced to go to intentionally underfunded public schoo.
• Environmental and resource concerns. You have but to reflect on the environmental impact to know, this is worrisome.

I love John Dolman’s appeal to all of us to embrace Epictetus or Stoic principles, which I’ll quote here for those unfamiliar:

Rejoice in what you have, cease worrying about things which are beyond the power of your will, make the best of what is in your power and take the rest as it comes.

Stoic Wisdom

I suspect that many teachers are wishing for the impossible, if they hope to go back to a time without AI and technology. Teachers, everyone who isn’t financially independent will find themselves forced to use AI. Only someone with only the most basic needs, a source of income, can hope to embrace Stoic wisdom:

“He is a wise man who does not grieve for the things which he has not, but rejoices for those which he has. Remember that it’s not only the desire for wealth and position that debases and subjugates us, but also the desire for peace, leisure, travel, and learning. It doesn’t matter what the external thing is, the value we place on it subjugates us to another . . . where our heart is set, there our impediment lies. Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things you only hoped for.” -Epictetus

A quick review of those Stoic habits John Dolman refers to:

My 7 Habits of Highly Stoic People

1. Be self-aware.
2. Control what you can, let go of the rest.
3. Practice gratitude, find joy in simplicity.
4. See obstacles as growth opportunities.
5. Live in the present, no past regrets or future worries.
6. Cultivate inner peace against chaos.
7. Strive for virtue and wisdom.

The more I blend AI into my workflow, the more I ask myself, “Are these benefits worth the price paid by others, the environment?” But then, that’s true of EVERY technology that powers modern society, no?

Ok, enough philosophy. Let’s get down to brass tacks for educators.

Two More Questions

Mike Bell, author of The Fundamentals of Teaching, suggests we ask a tough question when someone says, “The research says its effective.” In the case of AI, preliminary research is already suggesting it’s not effective for student learning. It stunts their brain growth in ways that hamper critical thinking and long-term information retention. More evidence is needed, of course.

Bell’s point still applies. He writes:

When someone says, “The research says it’s effective,” ask for the effect size. If it requires a lot of work to implement, and has a low effect-size, pause.

What is the effect size of AI in the classroom? Does it cost a lot of money?

But, of course, that’s the opposite of what’s going on with AI, right? Every company that can make money from AI adoption in K-16 education is promoting it. How can humanity overcome the engines of its own economy? And, of course, schools can’t. They are the tail on the dog, whipped this way and that by whatever catches the attention of a creature with a toddler’s attention span.

Issues Remain

Let’s be honest. AI makes work a lot easier, minimizing the “friction” of making, organizing, analyzing content . Students need that work to develop their own critical thinking and brain power. Adults focused on making money are willing to use AI. The ethical dilemmas posed by AI tools are massive, but no different than any other technology (see table of tech and ethical concerns) humans use to the detriment of stakeholders.

Those stakeholders are the workers, the environment, anyone who is exploited.

Some day, there will be a reckoning. And AI, like every other technology will reach a point it is no longer an issue or providing such a fantastic benefit (think of your smartphone), that it won’t matter.

It is an unpleasant truth, but the majority of people in society will use AI, no matter its negative effects. After all, I don’t see anyone making their own clothes a la Gandhi in an effort to stop production.

Do you?

Table of Tech and Ethical Concerns

Examples of Technologies in K-12: 1980s and Onward

Both tables below were generated by Claude.ai:

Consider this list:

Here are some examples of technology being deployed in K-12 schools, along with their effectiveness, ethical concerns, and resolutions:

In previous blog entries, I shared a list of AI fears. Did those fears have any evidence? This is part 4 where I share the AI fear and an AI’s response with AI evidence. Skeptical? Good. I won’t mention the problem with confirmation bias, only remind you of Trecek-King’s quote:

“We think we follow evidence to a conclusion. In reality, we come to our beliefs in irrational ways, then work backwards to find evidence to rationalize the belief.” -Melanie Trecek-King, ThinkingIsPower.com

So, with that in mind, here’s the fear and the AI evidence. In previous blogs, I relied on Perplexity and Claude AIs. In this one, I’ll use Meta.ai. (ugh).

Read the series here:

• AI Fears: Lacking Evidence, Part 1
• AI Fears: Lacking Evidence, Part 2
• AI Fears: Lacking Evidence, Part 3
• AI Fears: Lacking Evidence, Part 4

FEAR #3: AI REPLACES PEOPLE’S FUNCTIONS

One of the tough things to comprehend is that AI can replace the work trained professionals do. It can do it effortlessly under the right tutelage and with the right prompts. This means that people’s work product, what they get paid to produce (as opposed to being paid to reason) is easy to replace. With one person and an AI, I may get away with cutting a few other jobs.

It’s not hard to imagine AI robots eventually taking over physical tasks that are dull, repetitive, and which employ low-skilled workers. Worse, we can easily imagine AI using its own power to leapfrog from advancement to advancement to replace jobs that involve creation of some work product.

AI’s Response

Here is the information in markdown format:

Evidence of AI replacing people’s functions

=============================================

Automation of repetitive and routine tasks

• Studies have shown that AI and automation can already perform tasks such as:
  • Data entry
  • Bookkeeping
  • Customer service with high accuracy and speed.
• According to a McKinsey report, up to 45% of work activities can be automated using currently available technology.

AI-powered tools and systems

• AI-powered tools and systems are being developed and implemented in various industries, such as:
  • Healthcare
  • Finance
  • Transportation to perform tasks that were previously done by humans.
• For example, AI-powered diagnostic tools can:
  • Analyze medical images
  • Diagnose diseases with high accuracy, potentially replacing some tasks performed by radiologists and doctors.

Job displacement and economic disruption

• The rise of automation and AI has been linked to job displacement and economic disruption in various industries, such as:
  • Manufacturing
  • Transportation
• According to a report by the Economic Policy Institute, automation has been responsible for the loss of around 1.5 million manufacturing jobs in the United States between 2000 and 2017.

Advancements in AI capabilities

• AI systems are rapidly advancing in areas such as:
  • Natural language processing
  • Computer vision
  • Machine learning enabling them to perform complex tasks that were previously thought to be the exclusive domain of humans.
• For example, AI-powered language translation tools can now:
  • Translate languages in real-time, potentially replacing human translators in some contexts.

It’s important to note that while AI has the potential to automate and replace certain tasks and functions, it also has the potential to create new job opportunities and enhance productivity in various industries. The key is to ensure that workers have the skills and training needed to adapt to an increasingly automated workforce. You can copy and paste this into any markdown-compatible platform.

In a previous blog entry, I shared a list of AI fears. But, when I ran it through Trecek-King’s FLOATER acronym, the AI megaprompt suggested that while well-reasoned, the fears lacked evidence. Of course, it’s not that the fears lacked evidence, but that MY original post was missing evidence. This is part 3 where I share the AI fear and an AI’s response with AI evidence. Skeptical? Good.

Read the series here:

• AI Fears: Lacking Evidence, Part 1
• AI Fears: Lacking Evidence, Part 2
• AI Fears: Lacking Evidence, Part 3
• AI Fears: Lacking Evidence, Part 4

Confirmation Bias

One of the dangers of AI, and this blog post and subsequent ones may well prove it, is that one starts with the desired conclusion (fear of AI). Then, one uses AI to analyze low quality, incomplete evidence that supports that conclusion in a fit of motivated reasoning (a.k.a. confirmation bias being the main problem). Confirmation bias is simply coming to a conclusion then marshaling evidence and “motivated” reasoning to support that conclusion.

“We think we follow evidence to a conclusion. In reality, we come to our beliefs in irrational ways, then work backwards to find evidence to rationalize the belief.” -Melanie Trecek-King, ThinkingIsPower.com

So, with that in mind, here’s the fear and the AI evidence. In Part 2 of this blog entry, I relied on Perplexity.ai. In this one, I use Claude.ai.

Is the conclusion merited or not?

FEAR #2: AI REPLACES THE HUMAN REASONING PROCESS

AI will replace critical thinking and reasoning. Learning how to reason is hard work. Then, learning about logical fallacies, how to overcome our own bias, is hard. Reasonign involves following evidence to a logical conclusion. Melanie Trecek-King (Thinking is Power.com) describes it:

Evidence->Reasoning/Logic->Conclusion

AI makes it very easy to rationalize, or to select evidence to justify a conclusion. Rationalizing might be described as: Desired Conclusion -> Motivated Reasoning (confirmation bias) -> Incomplete, low quality evidence

AI makes it a simple matter to construct motivated reasoning with any evidence to reach the desired conclusion. But the fear with students using AI is that students NEVER develop a way of gathering evidence, figuring out what it says, then coming to a conclusion. The reason why is that writing has been the way that’s been done for a long time.

AI’s Response

There are valid concerns that AI language models like myself could potentially hinder the development of critical thinking and reasoning skills in students if not used thoughtfully. Here is some of the key evidence to consider:

1. Ease of rationalizing desired conclusions: As you note, AI makes it very easy for students to cherry-pick evidence to support whatever conclusion they want to reach. They can prompt the AI to generate arguments for a particular view, without doing the hard work of gathering and evaluating evidence objectively. This risks reinforcing confirmation bias.

2. Outsourcing the reasoning process: If students rely too heavily on AI to analyze information and construct arguments for them, they may not fully develop their own skills in logical reasoning, evaluating evidence quality, recognizing fallacies, etc. Learning to reason well requires practice.

3. Potential to skip evidence gathering: A worry is that students will jump straight to generating a full essay with AI based on their existing views, without first doing a thorough investigation of evidence on the topic. Sound reasoning has to start with comprehensive, high-quality evidence.

4. Impact on motivation to learn: Some fear students will be less motivated to put in the hard work to develop reasoning capabilities if they can take an “easy route” with AI. Writing has traditionally been key for building these skills.

However, I would argue that while these are real risks to mitigate, AI can also potentially support critical thinking if used as a tool in the right ways - for example, by helping students more easily see different viewpoints on an issue to analyze, or by critiquing gaps in reasoning or evidence in student writing.

The key is for educators to very intentionally guide AI use as part of the learning process, with the goal of strengthening (not replacing) each student’s own thinking and reasoning capabilities. Students still need to be taught and assessed on their information gathering, logical reasoning, and drawing sound conclusions from evidence. AI should be a supplement to this educational process, not a substitute for it.

Proper instruction, support and assessment will be essential to ensure the next generation develops strong critical thinking faculties even in an age of AI assistance. The human reasoning process remains paramount. But AI, used carefully as an educational aid rather than a crutch, has potential to help cultivate these crucial skills. Research and experimentation will be important to figure out the best approaches.