Perhaps you start your day by asking your Echo to provide your daily update. A new AI tool called Huxe now allows you to accomplish a similar goal but in multiple ways that you can personalize. Huxe is a new venture created by some ex-engineers who worked on Google’s NoteBookLM and the service is similar in some ways to the audio overview available through that product. Presently, Huxe is available for iOS and Android.
Huxe is a free service, but it only make sense that ad or subscription versions must be on the horizon.
There are basically two ways to use Huxe. The most basic functions allows you to select from designated topics and the tool than creates a sort of podcast with two speakers based on recent “news stories” associated with those topics. You can access this feed repeatedly in a day and the feed will be different. The podcast identifies you, remembers when you last connected, and provides basic weather information at the beginning of each session related to your area. One of the most unique feature is that you can intervene while the podcast is playing, ask a question, and the AI tool will pause to respond. The idea is that you can go deeper whenever something is presented that you want to know more about. More about this feature later.
The second basic and less well-promoted capability allows a user to create what amounts to their own personalized “station” based on a prompt rather than the selection of designated topics, request new “episodes” from this station at later dates, and share this station with other users of Huxe. I have read multiple descriptions of Huxe available through web searches and I want to concentrate here on this second approach.
The following image shows the screen image you encounter once you have set up your Huxe account. The “play button” stands ready to play what I have described as your “daily brief”. To get to the personally created stations and the option to create your own, you use the icon within the red box at the bottom of the image.
This link opens up access to “shows” developed by others with the button allowing you to create your own appearing in the bottom right (see red box in following image).
At this point, you have the opportunity to enter a prompt to generate the topic you want to follow. My prompt for this example was “How has AI been used to individualize instruction in K12 classrooms.”
Unlike the “Daily Brief” which changes each time you connect, the “Shows” are created once, but you can then generate additional episodes using your original prompt.
As I suggested earlier, one of the unique capabilities of Huxe’s AI audio shows is that you can interrupt and ask a question. In the following image, you can see the point at which I interrupted because I was interested in a study being described and wanted a citation so that I could read the original study. Huxe responded with the title of the study and I was able to search and find the study.
Summary:
Huxe is an AI tool allowing access to audio programs created as daily briefs or personal shows. The personal shows can be shared. A useful feature allows the listener to interrupt a program and ask a question which the AI tool will then stop to answer. I rely on full text primary sources for what I consider my serious work, but Huxe seems a great way to access a useful morning brief to listen to while I have my first cup of coffee.
A podcast I was listening to suggested that you ask what your commonly used AI tools can tell you about yourself. This suggestion was made in the context of a discussion of AI security, but to me it seemed something interesting to try. From time to time, I search for my name using a search engine just to see what I could find and this always seemed to turn up something unexpected, but it had not occurred to me to try the same thing with AI.
I use a ChatLLM from Abacus.AI, which is a bit less expensive than the $20 per month tools, but it offers the opportunity to apply multiple AI services, which seemed like an interesting way to experiment with this task (see the following image). I used Claude for this example.
I used this prompt: What do you know about the Mark Grabe living in Minnesota?
I don’t have access to private or comprehensive background files on individuals, but I can share what’s publicly visible online for someone named “Mark Grabe” who appears to be associated with Minnesota and let you infer if that’s the person you mean.
From public sources, there is a:
1. Retired academic / educational technologist
Several profiles and sites point to a retired academic named Mark Grabe, described as an educational psychologist and educational technologist:
A personal profile on Glasp lists him as a “Retired academic – educational psychologist, educational technologist” and shows that he is highly active in reading and highlighting articles about education, AI, note‑taking, and related topics, with thousands of highlights and pages saved over several years [Glaspprofile](https://glasp.co/j3h2xdri1p632t4w).
He has a Substack newsletter called “Mark’s Substack” described simply as “My personal Substack” [Substack](https://markgrabe.substack.com/). Topics there (from titles and tags) typically relate to education, technology, and broader commentary.
He has posted as a retired academic still writing about educational technology and mentions Minnesota (e.g., comments about snow, fall colors, and Twin Cities parks) on social media [Threadsprofile](https://www.threads.com/@markgrabe).
While these pages don’t all explicitly say “Minnesota” in the bio line, the content and posts clearly reference Minnesota weather and locations, strongly suggesting he lives in or around Minnesota.
2. Author of education / ed‑tech books
The same name appears as the author of multiple books on integrating technology into teaching and learning:
“Integrating Technology for Meaningful Learning”
“Integrating the Internet for Meaningful Learning” (often co‑authored with Cindy Grabe)
“Learning With Internet Tools: A Primer”, etc.
These are listed on his author page on Goodreads [Goodreads–MarkGrabe](https://www.goodreads.com/author/show/312712.MarkGrabe). The topics line up with:
Educational psychology
Classroom use of technology
Internet tools and instructional design
That matches the description of a long‑time education/ed‑tech professor who later retired but stayed active writing online.
3. Likely background (inferred, not private data)
Based on the combination of sources above, the public picture of “the” Mark Grabe tied to Minnesota is:
Retired academic, likely a former professor of educational psychology/educational technology.
Has written multiple textbooks and practitioner books for teachers on integrating technology and the internet into instruction.
Is still active blogging and writing online about:
Educational technology
AI and learning
Note‑taking, reading, and study tools
Broader social/technology themes
I don’t have access to protected records (e.g., address, employer HR info, phone numbers, or anything similar), and I can’t confirm that every online “Mark Grabe” is the same person, only that these public profiles look very likely to refer to the same academic/author who mentions Minnesota.
If you’re trying to figure out whether this is a specific Mark Grabe you know (e.g., a relative, professor, or colleague), tell me:
In what context you know him (university, school district, family, etc.), or
Any known details (field, approximate age, institution),
and I can help you cross‑reference public information more precisely while staying within privacy boundaries.
Reaction
This information is accurate and does not contain anything I would call private information. The links are accurate and take you to some, but not all of my most active online accounts. For example, I have been a blogger since 2002 and while this is recognized, there is no link. The same is the case with my Mastodon accounts. Other AI tools I tried did identify these sources.
What about images?
It then occurred to me that it might be possible to generate an image to accompany something I might write about this content. I switched to Nano Banana.
Prompt: What can you show me about Mark Grabe.
The AI responded with “I can show you a portrait of Mark Grabe”.
After viewing the image, I asked Nano Banana to place a copy of the book “Integrating Technology for Meaningful Learning” on the desk resulting in the following image.
I don’t think those who know me would recognize this individual as me, but many of the attributes – hair line, glasses, beard are perfect. The AI must have found examples of my appearance to come this close.
Posted inUncategorized|Comments Off on Ask AI About Yourself
In a previous post, I mentioned TwinMind as an interesting tool that transcribes an audio input and then uses AI to summarize. While the technology is impressive and seems to work very well, my comments were mostly focused on what this tool allows and what it may prevent.
I certainly encourage you to take a look at TwinMind. The tool is available as an app and as a Chrome extension, with the Chrome extension being more capable. There is a free plan and a $10 a month pro version with the current free plan being ideally suited to student use. TwinMind was created by some of the developers of NotebookLM who decided to create their own company (nice description of origins and capabilities). It has several nice features for those concerned about privacy and operates on your device. The transcription and summarization capabilities do not require that any data be uploaded for assistance from a corporate AI service. I can’t say I use this tool personally, but I have been exploring its capabilities by opening my iPad and launching the app while I participate on my desktop in my weekly two-hour book club. The transcript, complete with “you knows” and (uhs) as well as the AI summary of both our small talk and book comments, was accurate (see sample summary below).
Concerns
Here is my objection. This tool would be great for applications such as having a complete transcript of a meeting (with summary) and perhaps sharing class notes with students unable to attend (perhaps the prof would generate and post to the course management system). However, there is value in taking your own notes that is avoided by relying on AI notes.
This study of the value of taking your own notes goes back decades and I address this research which I first studied in the 1960s, in multiple posts (here is just one). Taking notes requires thinking – selecting, paraphrasing, writing – and this thinking contributes to understanding and retention. You can review notes self generated or AI generated, but to this point the benefits of one approach versus the other has yet to be investigated. The closest comparison I can think of might be studies that provided what were called expert or instructor notes (e.g., Kiewra, 1985) and produced results superior to what students produced when relying only on their own notes (although using both personal and instructor notes produced the best results). Compounding the complexity of this topic is the issue of short versus long term benefit. Expert or AI notes may provide immediate benefit, but limit the development of personal skills important to functioning outside of a classroom environment. To my knowledge the benefits of AI have yet to receive much in the way of attention from researchers using sound methodologies. Asking learners what they think of the experience is not the type of work I trust.
Taking digital notes
How learners take notes (on paper or using a digital device) has its own controversy. I don’t intend to get into this debate in this post, but I back digital because I believe the opportunity to search, share, and reprocess offers advantages over time (one post).
I prefer a specific tool that allows simultaneous recording of audio from a presentation and links it to notes taken by the learner. This requires a digital approach. By connected, I mean the text in the notes is linked to the audio so that after saving the combination, selecting any given notes plays that audio at the point that the learner began taking that note. Because of this connection, the learner can review the audio later if a note does not make sense. When a learner is having difficulty interpreting a specific part of a presentation, I tell them to simply record something like ???? in their notes so they know what to listen to later when they have more time to focus on the problematic content.
I have made this recommendation for some years and see that the apps I originally recommended still work on my iPad, but some that were free now require a subscription. Interestingly in reference to my earlier comments about AI, many of the options I recommended that now require a subscription include AI, allowing summarization and interaction as added features.
Services I have used to record audio and take notes:
Like many application of AI in educational settings, it is possible to identify both opportunities and concerns. Notions such as second brain (storage of notes, annotations, etc.) recognizes the limitations of human cognition and the opportunities to augment cognition in ways that improve performance.
So many concerns for AI activity by learners fall into a common theme. Learners may use AI to replace an activity intended to develop a specific learner skill. By completing the task without personally exercising the intended skill, learners avoid the purpose of the task – learning a skill. This is obvious when used to avoid a writing task, but applies to a broader set of educational assignments. Taking notes is not necessarily a skill educators focus on developing, but researchers have studied how learners benefit in detail and in this case the activity itself is not the focus, but rather the benefits on understanding and retention that are improved because of the process.
I am not claiming reviewing AI summaries has no value, but am concerned that avoiding the process of taking the notes may be detrimental. As I have already stated, researchers have yet to investigate this topic. Note taking research isolates two major components – note making and note reviewing. Obviously, AI notes still allow for note review and the research on note taking indicates it is the less impactful of the two interrelated activities.
If I could presently create the ideal tool, I probably would add the opportunity to take notes in real time to something like TwinMind. The learner would then have personal notes linked to some type of audio or text transcript and the AI summary as a second study opportunity.
Reference
Kiewra, K. A. (1985). Providing the Instructor’s Notes: An Effective Addition to Student Notetaking. Educational Psychologist, 20(1), 33–39. https://doi.org/10.1207/s15326985ep2001_5
Like many educators, I have begun to question the impact of AI tools on learners. It is not the potential of AI that concerns me; it is how individuals end up using AI as an easy substitute for what are intended to be ways to develop essential skills and personal knowledge. Many who read my comments and tutorials are interested in Personal Knowledge Management (PKM) and the development of what is called a Second Brain. Therefore, I thought I would describe how the challenges facing university and K12 educators also seem to be creeping into the PKM and Second Brain space.
Those captivated by concepts and strategies emphasized by PKM and Second Brain advocates are typically adult, independent learners who believe these ideas can contribute to personal understanding and productivity. I tend to translate specific practices using the concept of generative processing, which comes from my background as an educational psychologist. From this perspective, individuals learn from what might be called thinking, but how to do this is often not understood. Generative activities encourage and involve external tasks that tend to engage productive internal (cognitive) processes. So, for example, if the understanding and usefulness of new information is enhanced by relating new information to existing knowledge, such connections could happen without external encouragement or could be encouraged and given direction by asking a simple question: what is an example from your personal experience that seems to be an example of the concept XXX? Questions are an example of an external task heavily used in education. Other examples of common learning tactics include taking notes and writing in response to a prompt related to content to be learned.
PKM and Second Brain strategies emphasize similar generative processes. There is a heavy emphasis on notetaking, summarization using a personal perspective, revisiting and reworking existing notes, and physically linking personal notes immediately as recorded and over time. The Second Brain perspective emphasizes the external storage and delayed use of the products of such activities. The argument for a second brain identifies the limitations of human memory and cognitive processes. It proposes that external storage systems can be developed to support cognition and augment the limitations of the human information processing system. Each component (the human brain and the second brain) are intended to partner in ways that each functions in what that component does best.
The observation that led to this post and its title was my recognition that new digital AI tools offered me ways to develop a second brain with little use of my own cognition. For example, I can use Recall to provide a detailed summary of complex documents, I can use TwinMinds to create a complete transcription of an audio presentation and store both the presentation and a detailed summary, I can use Smart Connections to identify links among the ideas appearing in summaries, and I can collect all of these products using a tool such as Obsidian or Recall. I can then ask the AI within one of these collection tools to write a blog post. All of this processing has involved very little of my own thinking or analysis. I can be from input to second brain to output with very little participation from my own brain. Of course, I can simply ask an AI to write something using a prompt without even identifying specific inputs to the example tools I have identified.
What is lost in classrooms and in the potential options now available to independent learners is the development of specific skills (e.g., writing, note-taking, critical analysis) and the products the engagement of these skills produces (notes or extended written documents). It is easy enough to avoid the downsides through personal decisions, but ask educators how well trusting personal learner decision-making works. I can commit to reading the inputs that influence what I write, taking notes, and finding interesting connections among the notes I have stored over time. I do not operate under time pressures that require a product, and I am not rewarded for meeting such deadlines. I have developed insights into the challenges of our present circumstances and admit to being concerned about what learners will lose, impacting their futures and the collective future we all face.
I welcome your comments in response to my observations.
Posted inUncategorized|TaggedAI|Comments Off on AI allows the development of a second brain while ignoring the first
I have been reading Adam Grant’s Hidden Potential, and Grant referred to a research study about teacher effectiveness in introductory college courses that sent me off on a deep dive into what type of college teacher is most effective. Grant’s topic was really about whether one should connect with the most expert person possible as a mentor, but his example and conclusions were based on a study focused on higher education, which raised an issue and possible stereotype I am sensitive to as a retired professor. In a section titled “What those who can do can’t teach”, Grant highlighted a research study that suggested contingent faculty (full-time, nontenure track faculty) are more effective with beginning students than tenure-track faculty. Right away, I worry that his argument suggests researchers don’t care or can’t teach. This was not really his intent, but just his way of framing the issue (the section title) was sufficient to motivate my investigation.
Grant’s position on expertise and guidance was based on a research study I will describe in greater detail after I explain his conclusion. The study (Figlio and colleagues, 2015) showed that students in Introduction to Economics performed better with full-time teachers than with tenure-track teacher/researchers, as demonstrated by higher performance in the subsequent related course. Grant explained that greater expertise results in greater difficulty communicating about the basics. Experts sometimes lack insight into their own skills, which makes it challenging to communicate effectively. They may ignore important information they assume others have or no longer recognize as present in their own thinking. Their explanations jump around with side examples that lack connections novices need to understand and put the ideas they are receiving together. I admit that some of this hit a nerve, as I sometimes received criticism in my own “intro course” teaching for spending time telling stories that some students thought were off target. To me, the stories were concrete examples I used to illustrate the concepts and principles I presented in practice. I was trying to demonstrate relevance.
So, off I went to read the original study responsible for this argument. Figlio and colleagues, who are economists, studied the students of contingent and tenure-track instructors enrolled in the first economics course. Their dependent variable, which I found strange, was the performance of these students in the following relevant course. This was argued to reflect the “value added” of the quality of instruction in the preceding course. My explanation for this approach was that this was archival data. Without equivalent performance instruments to evaluate performance in the initial course, the grade in the next course was a better dependent variable. These researchers didn’t speculate about their results in the same way as Grant. It is easiest to explain their speculation without reference to their statistical data. They produced a figure charting the distribution of the value-added variable for the contingent and tenure-track faculty and the distributions were nearly identical over the top 75% of both groups, but were different with fewer lower-performing contingent faculty with smaller value-added scores. These authors speculated that over time, more contingent faculty members who were less effective were rehired. I don’t know whether this explains why they were less effective so perhaps both sources are compatible.
How might administrators respond to these findings? Should they at this point? I guess my thought is that much more work is needed before decisions are made about staffing. First, just which institutions should be informed by these findings? The study was performed in one department of one institution. The institution, Northwestern, has a mixed mission including both research and instruction. The issue of tenure is likely heavily influenced by research and grant activity at this institution, but tenure as a practice is also what happens at many institutions with a primary instructional mission. The tenure vs. ongoing reappointment difference is a different issue.
Second, the economic focus on estimating value, as demonstrated by future benefit, is interesting but I think unusual when it comes to evaluating instructional skill. Most educational research focused on the impact of a practice examines achievement in the course where the manipulation or observed assumed causal variable is studied. The gold standard for such research (e.g. see Stroebe, 2020) would involve multiple sections of the same course, using a common method of evaluation, taught by different instructors, with students assigned to a section at random. This methodology is difficult to arrange, but it has been used often in addressing another critical question for administrators – whether student ratings predict instructional quality (the answer is no). When it comes to student achievement and whether instruction is provided by tenure track or contingent faculty using this method, I found one example (Carrell and West, 2010) that demonstrated better immediate achievement, but lower future performance associated with contingent faculty. So, there is more to be learned from additional studies.
Personal comment: I commonly taught the Introduction to Psychology course and I have considered the Grant argument about expertise as it applies to my experience. Introduction to Psychology is a survey course offered to both majors and nonmajors. Suppose you want to understand the content of such courses. In that case, you might imagine a typical textbook that contains chapters on such subjects as Research Methods in Psychology, Biological Bases of Behaviors, Learning, Memory, Sensation and Perception, Developmental Psychology, Social Psychology, Abnormal Behavior, Therapies and Treatments. These chapter titles could also be used for full courses available to students later in their studies. My expertise and the advanced courses I taught were in Educational Psychology, so Learning, Memory, Developmental, and Motivation were areas in which I propose I had some expertise. Despite the typical reaction when I am introduced to others as a psychologist, that I must be a clinician treating those with mental disorders, I did not study these areas and knew far less about them than graduate students with that focus in our own department. What would Grant’s argument suggest about my instruction in various topics covered in Intro, given the great variability in my actual expertise? I would suggest that pedagogical skills and content knowledge are both critical, yet somewhat independent, in impacting student learning. I have no data to evaluate my own contributions, but I would bet I provided a superior experience in explaining human learning compared to human pathology.
These are interesting issues with real consequences. Some suggest that administrators seriously consider hiring for two categories of faculty members in multi-mission institutions. I think there are related problems with this approach such as prestige and whether those focused exclusively on teachers are actually prepared for this role.
Why do I present this information? I suppose I assume some are just interested. I also propose that public perceptions of higher education are often poorly informed and I hope that indicating those involved are committed to exploring multiple productivity issues will help.
References:
Carrell, S. and West, J.(2010). Does professor quality matter? Evidence from random assignment of students to professors.” Journal of Political Economy. 118(3), 409-432.
Figlio,D., Schapiro, M. and Soter, K. (2015). Are Tenure Track Professors Better Teachers? The Review of Economics and Statistics, 97(4), 715–724.
Stroebe, W. (2020). Student evaluations of teaching encourages poor teaching and contributes to grade inflation: A theoretical and empirical analysis. Basic and applied social psychology, 42(4), 276-294.
These are the photos I was looking for, and this is the story of how I used the new AI feature in Google Photos to find them.
My wife and I are about to set out on a month-long adventure that will take us to Alaska and then Japan. The strange combination is due to what is called a boat “repositioning” by the cruise industry. This is the fifth trip we have made to Alaska starting with driving the ALCAN highway when some of it was still gravel, several ferry trips through the inside passage, and now a cruise ship.
As I prepared for this trip, I tried to locate images from past visits. I have always taken photos and written about our trips. Still, this content was collected over nearly fifty years and was created with multiple cameras and phones and scattered across online storage, numerous computers, and various types of storage media. When I retired, I undertook a project to collect as many images as possible in one location. I chose Google Photos, mainly because the images would automatically upload as I found them.
My situation is not unusual for someone of my age. If you are an older individual, you already probably know that your kids don’t want your stuff. They also are not that interested in your pictures. Even those in which they appear. This does not mean that you won’t appreciate the opportunity to review past experiences. I find that selecting a few and generating the stories that go with them is more likely to generate interest in others.
Back to the bears.
During one of our Alaska trips, we visited our daughter and her future husband, who were spending the summer exploring and working on a whale watching boat. The daughter was in a phase where she took a lot of photos and even blogged about her experiences. We were walking on the shore and she was using my camera when we saw several bears, one of which was out in the water. She captured a series of images as the bear moved through the water and at one point, climbed and posed on a rock before moving on. The bear on the rock picture garnered significant attention and appeared on the front page of the Ketchikan paper. I had the famous photo, but not the sequence, and that is what I wanted to see if I could find.
Here is the challenge. My collection includes thousands and thousands of photos. Over the years, I have paid for and maintained a collection in Flickr that is made up of more than 7000. The issue with Flickr was that I had used it for many years to showcase my high-quality shots. I was not using the service to collect the large number of images I deemed of lesser interest to others. The other images accumulated in iPhoto on several machines and perhaps as slides or digital photos stored on CDs or whatever happened to be the external media of the day. As I rediscovered these images, they ended up in Google Photos in no particular order. Note, few of these images had geolocation data, and many not even the EXIF data that would indicate the date the images were taken. Cameras, an SLR in this case, do not have the means to record the geolocation data we expect from cellphones.
Here is the feature Google has added to photos that seems added just for my situation. You can now ask Gemini (AI), available within Photos, to perform a wide range of tasks based on your photo collection. First, the Ask feature does not presently work on a Macintosh, but does work with the Google Photos app on the iPhone or iPad. Ask appears at the bottom of the screen instead of the search icon, which may be confusing if you are used to the traditional search method. Entering the prompt “Find images of bears in my photo collection” creates a long list of hits – bears in zoos, a photo of a black bear carved from a tree stump, teddy bears being held by our kids, and all of the bear images taken in Alaska. I was able to locate one of the missing bear images as the bear approached the rock I could not find in any other way.
I tried one more search that also produced some very surprising results. I knew that we had fished for salmon on one of our Alaska trips and searched for salmon. I found photos I had taken of salmon in restaurants, pictures of smoked salmon within a display case I took while visiting the Upper Peninsula of Michigan, and pictures of salmon taken while fishing in Alaska. The interesting thing about the fishing pictures was that these images appeared not as photos, but within videos. No tags, no meaningful file labels – AI found the salmon within the videos.
How can this work? The only approach that seems reasonable is that Google “processes” stored content assigning what I might called tags to things and actions associated with blocks in files and frames within videos. A prompt then uses these existing tags to generate a response. So, I am speculating that Google preprocesses the content you give it for storage much like say the content you upload for use in NotebookLM.
Summary
Google has added AI to specific implementations of Google Photos. The AI allows intelligent search that is not dependent on stored text, dates, or geolocation data. The AI can use these sources, but can also identify imagery directly. This capability is not presently available on all devices, but works via a device that makes use of apps (iOS or Android).
Posted inUncategorized|Comments Off on Google Adds AI to Google Photos
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.
You can revoke your consent any time using the Revoke consent button.
You must be logged in to post a comment.