The Curmudgeon Speaks

I acknowledge that I temporarily violated the Huxe Terms of Service in a recent post. Yes, I failed to read the TOS when I set up my account, but this is hardly uncommon. My violation was posting a link to content I created, even though sharing is not allowed.

Huxe is a new tool that enables AI-generated podcasts. At least this would be one way to describe the product. You identify sources or offer a prompt that sets a goal and you then have access to podcasts based on these designations. In addition, Huxe allows a user to interrupt a podcast and ask a question about the content – a very cool feature. 

My attempt to include a link to my creation, and my intent to record the audio as an example for those who had not created an account, was quite innocent. I made assumptions based on what I saw on my screen. There was a way to change my creation from private to public and the button to share the link to the creation. How would you interpret these signals? 

I discovered my transgression when I could not remember the process I had used to share my example and asked some AI sources to describe the sharing process. The second effort brought a response that sharing of this material was not allowed.

The first effort, which was a Google search that returned its AI summary before the links, not only indicated that sharing was legal but also provided a source for this claim. I checked the link, and ironically, the source was my original post, which included the offending link.

A couple of thoughts, I had always assumed that AI-generated content, being a summarization, was not protected by copyright. Perhaps the Huxe expectation is based on the “use of” rather than on the content itself. You cannot access the content without using the service, which seems to be the justification. I do think it somewhat ironic that, in comparison, the use of my content in a response created by AI without my consent would be acceptable. Something seems off in this combination.

I removed the shared example to resolve the issue, but the claim in Google’s response to my query did not change. This adjustment may take some time.

Actions taken in response to the recent tragedy involving the murder of National Guardsman (actually a woman) by an immigrant from Afghanistan fits the classic problem of anecdotal thinking. This limitation in human cognition is common, but the manipulation of humans because of this limitation can be quite sinister.

Trump’s reaction to this tragedy reminds me of an example of flawed thinking commonly taught in Introduction to Psychology. I call it “anecdotal reasoning”. It is something we all do in daily life when we use an example to illustrate a claim we want to make. The problem with anecdotal reasoning is that it presents a limitation of generalization. Many of the claims about immigrants fit this problem and the present case represents a perfect example. The anecdote is atypical. Immigrant crime is lower in general than crime committed by citizens. However this is explained – immigrants have more to lose, immigrants are better people, etc. – is not relevant here. The fact is this is a fact. To make a case for a general strategy using a limited number of cases or a single case is not logical and if done purposefully because people easily fall for this weak type of thinking, manipulative and in some cases immoral. Muslims, those practicing or simply those from Muslim countries, do come from a different culture, but differences are also true of Americans in general. Rather than see this tragedy as a sign of some weird sleeper cell espionage conspiracy, start from the most likely scenario until proven otherwise. This seems to me to be a suicide by cop attack based on mental health issues or perceptions of betrayal and powerlessness. If you and yours are lucky enough to avoid these life challenges, you are not special, you are privileged.

Concerns about AI often seem abstract and distant. This seems the case with AI and job loss. We understand the reality, but we lack the personal connections that are often required to motivate people to consider and pursue related actions.

I have realized I am a participant in the problem if only as a consumer. I have been a blogger for years and at some point I was told that you should include some type of image in your posts to make them more interesting. I have limited artistic ability beyond photography so to include some type of imagery in some of my posts I resorted to purchasing clipart. I used a service called the Noun Project. At the time, I think I paid $20 a year to select images from the giant Noun Project database and use them in my posts. The company name indicated how the process worked. You would enter a simple request and this request would bring up options you might select. For example, I might enter “classroom computers” to find examples the creators had tagged with that phrase. An example from one of my old posts follows. For years, I used such images in my posts and classroom lecture slides. Those contributing images received a small amount each time one of their images was selected.

Now with AI tools (e.g., Nano Banana) you don’t see many graphics of this type anymore. You can access such tools for no cost if you are not a heavy user or as part of a paid account if you are a constant AI user. The sophistication and flexibility is so much greater and the images are far superior (see following image from a recent post). 

Somewhere, there are graphic artists who no longer have the income the Noun Project provided. 

I saw these swans while driving yesterday. They were confined to feed in a small area of open water surrounded by ice. I thought swans migrated and my Internet search says they do when necessary. It looks like necessary is closing in.

A frequent complaint on social media is critical of higher education and urges people to support spending more of the money the government sends to higher ed on tech schools. These messages are misguided because the grant money that catches the public’s attention is typically focused on research, rather than tuition. Aside from this misunderstanding, I react to the claim that higher ed is not about practical or productive things.

This time of year we often visit apple orchards. When in the Twin Cities, our home, we try to make at least one trip to the University of Minnesota Apple Store. The U of M is a land-grant institution and has agriculture as a core mission. As an urban campus, this might surprise some citizens and even students because the main campus and the ag campus are in different locations (different twin cities).

My own education was completed at a land grant university (Iowa State), and I grew up on a farm, so I have some affection for this mission. I admit I had not really appreciated this role at the U of M until we moved here. Agriculture can involve many different things, and agricultural colleges often have unexpected areas of specialization. Among the specializations at the U of M is horticulture and, more specifically, apples. The Honeycrisp apple, possibly the most popular apple in production now, was developed here.

No, these apples are not sold here at bargain prices, but we were able to purchase slightly imperfect ones at a lower price.

I always wonder what it would be like to take classes in that department.

I admit that I find AI search summaries useful and read them first when they are available. However, perhaps because I have several blogs of my own, I can’t help wondering what are the long term consequences for the original content creators who author the content feeding AI search summaries? What are the incentives when the AI summary frequently serves as the alternative to reading what they have written and recognizing their contribution in some way?

Like so many uses of AI, search summaries aren’t inherently harmful. A summary is just that and is different than the unique information you would get from a single source. I don’t know that this is the case, but it seems a summary would be less biased than a single source. However, we risk a web where knowledge is consumed without context, creators aren’t rewarded, and the public sphere narrows. 

Yes, I know at least the Google summaries provide links to at least some of the original sources used within the summary. I had a feeling based on my own approach that many of these links would be ignored. When I write, I do like to acknowledge my sources. This seems the fair thing to do, but is also a way of increasing the credibility of what you have to say as readers can review the original material. This is a practice I retain from my academic training. However, I also admit, I often visit a single source and if it is appropriate to back the data or fact I want to report, I retain only that citation.

I assumed that there would be quantitative data on the use of summary citations even if the companies offering the summaries do not offer this information. The data points I include in the following material come from the PEW Research Center. These researchers found that users who encountered an AI summary clicked a traditional result only 8% of the time, compared with 15% when no summary appeared. This means users were less likely to use both the summary and the other search results. They were also more likely to end their session entirely – 26% vs. 16% without the summary. If a summary was available, further searches were less likely to follow. This was argued to suggest the overview often functions as a stopping point rather than a starting point for exploration. Even links inside the AI module get little traction and are not explored in detail. Those who did clicks occurred in just 1% of visits to pages with a summary. This is the counter argument to the argument users will use a summary with citations as a starting point. 

The same analysis indicated that some sources appear commonly in summaries. Several examples were provided with Wikipedia and YouTube among the most common. This makes sense as both are what people might have searched before search summaries were a thing, but also begs the question of why not go directly to Wikipedia. I suppose is the answer is you would still have to read the Wikipedia article. 

We seem to be moving toward a situation in which what used to be search and what we first encountered and continue to encounter with AI tools are little different. You enter a prompt and review the response.

I have been thinking about what sources might be most harmed by this emerging reality. I think news sources will fare the best because many of us use a news source for immediate information and far less as a valuable archive. With the decline in the use of opportunities such as RSS, I find that most of the hits I get on my own content comes from search and the more popular posts accumulate reads gradually over time and not abruptly when first completed. 

Is this just sour grapes on my part? Maybe, but I do wonder about the long term consequences of this pattern. When I think about the motivation of writers, especially those of us in the long tail, this is seldom about money. It is most about the attention and the satisfaction of knowing someone has read YOUR ideas. 

I wonder if content aggregators won’t become more popular if they protect content from AI access. This would seem attractive even if content creators paid a bit for this protection. The attitude that there is no difference between an individual and an AI bot summarizing your content appeals to some and an approach like this might require legal protection, but this is the only approach I can imagine that seems productive.

I recently had what I thought was a very amusing experience, my wife thought offered an insight into digital natives, and I categorized it as an example of a First World Problem.

I was at my favorite coffee shop and was about to leave when I encountered a twenty-something walking along and looking beneath the row of tables along the wall where I was seated. I guessed that he was looking for an outlet to plug in a digital device and said that I was leaving and he could have my table. He then explained that his phone had died and he needed to get at least enough charge into his device so he could unlock his Tesla. In the first reaction, this may seem unusual, but it is logical.

I rose to leave and then watched as he stared at the outlet next to my table, realizing that he had a cord attached to his phone, but there was no converter at the end of his cord, and the outlet would be of no use to him. Trying to help, I pulled my laptop charger and cord out of my bag and said that while I had to go, I could give him five minutes. He said that should do it. After about five minutes, he thanked me and left. This was such an unusual experience I have since thought a lot about it. Maybe it is just my perspective as an old man, but this one still has me shaking my head.

I found great value in the “No Kings” protest. Despite the MAGA characterization of a “hate protest”, I found the large local gathering anything but about hate. We are in a time of great and dangerous disruption of how our government should function. Given that we are less than a year into the Trump presidency and several weeks into a government shutdown, it is easy to become dispondent and feel helpless. Such large gatherings offer the perspective that your reactions to the present circumstances are far from out of touch, and demonstrate that there are many like you out there willing to make the effort to take to the streets in support of a more positive and less selfish view of our country. Participate in future events of this type. It is a great and positive experience.

I have just finally gotten around to reading Jaron Lanier’s book, Who Owns the Future, and it has given me much to think about. It has been the most recent focus of the book club I participate in, chosen because we read extensively about both politics and digital media. Lanier argues that the increasing focus on digital networks, while initially quite positive, has shifted to take wealth and influence away from the middle class, contributing to a general recession. While many of us contribute content and attention to create a meaningful experience, the central computer companies make nearly all of the money. If you are not paying for the product, you are the product is often used as an explanation for what seems to be free services. Sometimes, things go even further, and you both contribute content and pay to do so. 

Lanier identifies a common pattern across what he describes as Siren Servers. He uses the term “Siren Servers” to describe large, centralized computing platforms that lure everyone into using “free” services, vacuum up data at a massive scale, analyze it with superior computer power and algorithms, and then concentrate wealth and power in the hands of the few who control those servers. There is a certain “bait and switch” element, intended or perhaps not, in what tends to happen. Users are attracted and invest time in various ways and these investments result in a network effect that traps them in various ways. They may build connections with others who are loosely joined, offering advantages, but they cannot be easily moved to a different network. They may invest in creating content that has aggregate value, but might also be challenging to move. The switch occurs as those behind the server act to “decouple risk” so that it moves to those using the system and is not shared with the system. 

By chance, I happened across a Medium article, “Medium writing platform exhausts its potential by eliminating incentives from subscriptions …” by Jamen Mendes. The similarities to what Lanier described immediately struck me. Mendes notes that during its origin, Medium shared the monthly subscription fees with content creators and members, based on how much paid subscribers engaged with individual authors’ work. Medium gradually shifted away from this model to completely abandon sharing subscribers’ subscriptions in favor of ad revenue provided through individuals or organizations willing to pay to increase the degree to which their posts would appear in the stream of titles sent to users and reader activities other than reading (claps and interactive comments). Mendes contrasts the present Medium system with Substack, which uses a more direct relationship between subscribers and author compensation. 

As a content creator (aren’t all social media users), this is the first time I have considered some of these issues. Obviously, there are differences between the companies Lanier writes about and Medium. Medium is tiny compared to the corporations Lanier targets, which rely on collecting and selling information to ad companies and others interested in user behavior data. Medium is not a player in manipulating political matters or the economy. However, there are parallels in what Lanier describes as risk. Like other Medium subscribers over the years, I acknowledge that the changes made were not done without acknowledgment, but the rationale was never clear to me. I understand that the backend necessary to support micropayments models requires storage and compute, but why was this not adequately incorporated into the original business model? Lanier’s perception of shared risk somehow converting to user risk does seem to apply. 

I apologize in advance for this post, but I quite frankly have had enough. If it matters, I posted this would required membership to read because I understand this content is the typical effort I make to offer content I think is potentially educational.

I doubt if is is surprising to any regular Internet user that you encounter plenty of social media content you regard as objectionable and unfair. Mostly, I simply scroll through this material without much of a reaction. I am having trouble doing doing this with criticisms of those I regard as experts in ways that offer nothing to support the claims made limiting any opportunity to challenge the claims made. I encountered the following pronouncement which is typical of the type of comment that annoys me.

I was a college prof some years ago and also had opportunities to observe K12 classrooms. I don’t remember ever witnessing the type of objectionable behavior hinted at here. Maybe times have changed with everyone functioning in a more hostile environment and even encouraging more educators to comment on current issues. I doubt this is the case, but I admit it is a possibility. 

The type of ambiguous complaint that is the focus of this Facebook post has become much more common. Immediately, I react because there is nothing specific about the complaint or even an assurance that the complainer was a witness to inappropriate behavior. 

I imagine educators reading this trash and not feeling like there is a reasonable way to reply. Most educators do not teach subjects that would naturally bring up the type of content likely to be offensive – maybe social studies or English composition. I suppose the biology teacher might explain evolution or maybe something related to the biological definition of male and female and perhaps the difference between the genetics and concepts such as gender identity. Since the accusations are so often so vague I have to work hard to imagine exactly what topics or comments might rise to the level encouraging parents to react.

Frankly, it seems of greater potential concern that parents poison their kids understanding of many now politicized issues. What these kids now believe can make situations difficult for educators and for other students. What does the teacher do when positions are taken in class when arguments are advanced that are simply wrong or clearly not in keeping cultural norms.

I understand the issue of anecdotal reasoning. Someone has had a personal experience or has an acquaintance who has had an experience that is rare, but this experience does fit the position being advocated. When people are emotionally involved such evidence may be sufficient, but in rational situations most of us can take rare events into consideration. 

Empathy is a part of this problem. Have the complainers considered the impact their claims could have on others again without clear identification of problem, circumstances, or the actual target of the accusation? 

I have little to offer in terms of advise. I sometimes take the time to respond based on the issues raised here. What specific behavior has prompted this attack? How often has it happened? Did you witness the type of issue you have described or is your position based on other nonspecific complaints you have witnessed on this platform? I have no idea if such responses impact the poster, I do want those attacked to understand others did not support the claim being made.