Zak Kolar

In response to Mekka Okereke:

The 10 women might answer the “Can people from all backgrounds succeed here” question with an answer of only 50% positive.

The 100 men, after observing the women’s experience, usually lower their scores.🤯 They might now answer the question with an average score of 70% positive.

This is a good thing!

The world that exists in the minds of your men employees, and the one that exists in the minds of your women employees, are coming closer into alignment with the way the world actually exists.

(Full thread)

This reminds me of “A Nation At Risk”, the Reagan-era report about the state of American education. It lamented lower average SAT scores, which was true but misleading. At the time, the college-bound population was growing. More students who had generally been excluded (including women, Black people, people from low-income backgrounds) were able to attend. This means the pool of SAT test-takers was bigger and more diverse than in prior years.

The dip reflected not a decline in the overall quality of education, but a more holistic picture of the landscape at the time. When the scores were separated by subgroup, most groups' averages had actually improved from previous years (Simpson’s paradox). That said, this exposed long-lived inequities between the opportunities afforded to these groups that needed (and still need) to be fixed.

With more nuance in reporting and interpretations, the country could have addressed systemic inequities that disproportionately limit opportunities available to Black, Hispanic, and Indigenous children. Instead, the one-size-fits-all approach has perpetuated them through policies like No Child Left Behind, Race to the Top, and the general obsession with high-stakes testing.

View on Mastodon: Part 1 Part 2 Part 3

Every few months, my iPhone’s keyboard seems to degrade. More mistakes slip past autocorrect, and I have to type words 3-4 times in a row to remove typos. Sometimes, it even adds its own typos when I get a word right.

My theory is I rely on autocorrect too much, and that creates a feedback loop over time. I tend to make the same mistakes, like tapping the space bar too close to the N and getting phrases like “I amngoingnto”. In the beginning, autocorrect assumes this is a mistake and fixes it. But as my sloppy typing turns into muscle memory, the keyboard’s machine learning algorithm treats the frequency as intentional and adjusts accordingly. It starts ignoring my most frequent stray Ns and even replaces spaces with Ns.

Whatever the reason, I’ve found resetting my keyboard dictionary when these issues arise tends to help. To do this on iOS, go to Settings > General > Transfer or Reset > Reset > Reset Keyboard Dictionary. These taps get progressively scarier - I’m always afraid I’ll accidentally reset the entire phone. But, at least in iOS 17, the Reset button triggers another prompt to choose specifically what to reset.

I go through this process every few months. There’s a small adjustment period in the beginning while the keyboard re-learns my most frequently used words, but this is much less annoying than fighting the bad habits it picks up over time.

Finished reading: Digital Minimalism by Cal Newport 📚

This is a great complement to Stolen Focus by Johann Hari. It has strategies to help reprioritize your time, attention, and relationship to technology. It’s not an all-or-nothing approach - you can pick out which approaches will work for you and skip over the ones that may be less relevant or beneficial.

These are some ideas I’m trying, some directly taken from the book and others indirectly inspired.

Thinking time

I’m setting aside 30 minutes each evening to just sit without my phone or any content to consume (TV, books, etc). I’ve decided to allow music as long as it’s instrumental or I can tune out the lyrics. While the goal of this time isn’t explicitly to write, I’ve already filled a few notebook pages with thoughts ideas from doing this the last few nights.

Apple Watch

I’ve stopped wearing my Apple Watch during the day. I’ve already noticed a huge improvement in concentration without getting disrupted by notifications the last few days. Since I’m so used to the watch, I’m not in the habit of checking my phone frequently without being prompted. I want to avoid falling into that when the novelty wears off.

Screen time limits

Mostly for privacy purposes, I deleted many social media apps from my phone several years back and switched to the mobile websites when I wanted to check them. I use the Duck Duck Go browser for social media sites. This was another privacy move initially, but an added benefit is that it signs me out each time I clear my tabs so I need to sign back into my accounts each time. This adds just enough friction to prevent mindless scrolling. It’s also reduced my impulse Amazon purchses.

All that said, I’ve found other apps to get sucked into that are more privacy-friendly (Readwise Reader, Mona for Mastodon, micro.blog). I’m going back and forth between setting screen time limits and deleting some of those apps as well.

Conversation Office Hours

One of my favorite suggestions from the book is to create “conversation office hours”. This is a dedicated recurring time when people can call you (or drop by) to have casual, yet deeper interactions than social media comments or even texting provide. It reduces the anxiety and awkwardness around calling out of the blue because callers know they aren’t interrupting you. At the same time, it’s organic and flexible because you aren’t scheduling specific times with specific people.

One example in the book is a person whose “office hours” are during his evening commute. His friends and family know they can call any weekday at 5:30 and he’ll be available. My commute is (thankfully) short, but I could use the time when I prep/cook dinner. Turning this into a routine might even help avoid the temptation to order out, which is another goal of mine.

Experimenting on myself – Mike Crittenden

I can convince myself to stick with anything if I tell myself it’s only a week-long experiment. For the past 6 weeks, each Monday morning I’ve settled on a new weekly experiment

I have so many things I’ve planned to start “next week” (eating better, going to the gym, writing). A week-long experiment sounds like a way to get out of the infinite “next week” loop.

Cory Doctorow - What kind of bubble is AI?

Just take one step back and look at the hype through this lens. All the big, exciting uses for AI are either low-dollar (helping kids cheat on their homework, generating stock art for bottom-feeding publications) or high-stakes and fault-intolerant (self-driving cars, radiology, hiring, etc.).

Keeping up with AI developments can feel like drinking out of a fire hose, so many educators are wrestling with this question. There is no one-size-fits-all answer, but my hope in this post is to provide background and suggestions as you navigate the flood of information to help separate hype from practicality.

Large Language Models

Artificial Intelligence (AI) is a broad term without one single agreed-upon meaning. Large Language Models (LLMs) are one type of generative AI that use large data sets of text to generate new text, such as ChatGPT. Many platforms marketed towards educators utilize LLMs, and this post will focus primarily on those.

Considerations

Before deep-diving into AI platforms, here are some considerations to keep in mind.

Student Privacy

Most, if not all, AI platforms have an age requirement that excludes students below age 13, 16, or 18. Many platforms also train their models with user-provided data. This means providing any student information or work, even if the student isn’t directly interacting with the platform, may be a violation of FERPA regardless of student age. Once input has been used to train the model, districts no longer have ownership over it and it cannot be retrieved or deleted should a student or guardian make the request. Input used to train the model may also make its way into responses that are generated for other users. Several companies, including Apple, Goldman Sachs, and Samsung, have banned employees from using ChatGPT, citing the potential for confidential information to be leaked among other security concerns.

Accuracy

LLMs are mathematical models that ingest huge amounts of writing, mostly scraped from the internet, and generate combinations of words that are statistically likely to appear together. When asked to complete the sentence “The school is…”, the LLM will guess which word is most likely to come next based on similar combinations of words in its training data. It has no concept of the meaning of “school” and therefore is incapable of determining whether the word it picks is accurate.

Originality

Because LLMs generate their responses based on existing writing, they cannot produce new ideas. They may synthesize multiple sources at a superficial level, but they are not capable of analysis or original thought. At best, they provide the “average” response a human is likely to write based on their training data. This “average” is based on frequency in the training data, not accuracy.

Attribution

LLMs can’t attribute ideas to their original sources. Because they are stringing together phrases and sentences using statistics, they have no concept of the original source of an idea within their training data. When asked to provide sources, they often write citations that are relevant to the given topic but aren’t necessarily the source of the information to which they are attributed. In some cases, they create realistic-looking citations to sources that don’t actually exist.

Pricing

Right now, many AI companies are subsidizing the cost of these tools for research and/or marketing purposes. LLMs cost companies hundreds of thousands, if not millions, of dollars per month, so keeping them free forever is not sustainable. It’s likely that in the near future, these companies will go out of business before they become profitable or start charging for their products.

AI Detection

As of this writing, there is no reliable way to determine if text has been generated by a LLM. OpenAI, the company behind ChatGPT, has discontinued its own tool citing accuracy issues.

Third party tools are subject to the same problems. False positives harm students' academic careers, and work written by non-native English speakers is disproportionately flagged as AI-generated.

Impacts and Implications of AI

If you’re looking to engage your students in critical thinking about AI, Automm Caines' Prior to (or instead of) using ChatGPT with your students outlines activities and discussions that examine the impacts and implications of AI and other technologies. You can use these to expose students to these tools in a way that doesn’t jeopardize their privacy and intellectual property.

Further reading

• What is Academic Integrity in the Era of Generative Artificial Intelligence? by Chris Dede
• Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell