Study Findings Mount: Human Texting Outperforms AI Chatbots in Alleviating Loneliness

"Texting a stranger is better for reducing loneliness than an AI chatbot, study finds" From CTV News: https://www.ctvnews.ca/vancouver/article/texting-a-stranger-is-better-for-reducing-loneliness-than-an-ai-chatbot-study-finds/ "The study found that only those who texted with a fellow human reported feeling less lonely at the end of the experiment. 'We thought that interacting with AI might be as helpful as texting with a random fellow first-year student,' said the study’s author, psychology PhD candidate Ruo-Ning Li, in a media release. 'But to our surprise, only the human-to-human texting reduced loneliness over time. The chatbot, even though we designed it to be the ideal supportive friend, didn’t shift loneliness.'" #NoAI #Loneliness

While not therapy, talking to literally anyone is better than talking to an LLM. Texting a Random Stranger Better for Loneliness Than Talking to a Chatbot, Study Shows https://www.404media.co/chatgpt-loneliness-study-college-students-random-strangers-texting/ #Texting #Loneliness #Chatbot #AI #MentalHealth #Tech

"This is just such a low tech, simple intervention, and can make people feel significantly less lonely." https://www.404media.co/chatgpt-loneliness-study-college-students-random-strangers-texting/ #study #MentalHealth #loneliness #ChatBots #AI #LLMs

Lots of people who strongly dislike #AI here, for all kinds of reasons. I have some weird idea that I'd like to bounce off of folks - no wrong answers, I'm really just curious if this is something that could or should be pursued or if I better stash that on the pile where I keep all my other stupid ideas. Thing is, these chat bots look somewhat useful on a purely "when putting stuff in, somewhat reasonable other stuff comes out" basis. There are tons of downsides to the technology, among others: centralized control, high hardware costs, the ecocide that comes with how these central huge deployments are operated, the disenfranchising nature of the perception of "specialists no longer needed". Now, the last point is different from the others in my opinion: If that stuff doesn't crash-land hard¹, we'll have to deal with it somehow - and I wonder if the best remedy against a non-specialist trying to get rid of the specialists by using a tool isn't equipping the specialist with that tool, so they can wield it better than the non-specialist (while still giving the non-specialist better tools to "scratch their own itch", as the foundational writings of Open Source like to say) Which leaves those other issues, which mainly come down to "the tool is built for industrial scale use, and there's no EPA to reign it in." So, my fancy is computer science, with a focus on coding, so that's what I'll look into here. The concept might be applicable elsewhere. For the purpose of using LLMs as a coding aid, what do y'all think if they were used like this: 1. Choose/Build (as a community, perhaps) some base model with proper rules on data provenance (e.g. only attribution-only licenses, and accompanied by a document that lists everything that went into it, which can serve as CREDITS file for changes built with the support of the model), built with only green energy, etc. Keep the model small, but not too small. It would be reasonably powerful (not industrial-scale powerful) but probably too large for most folks to run at home. 2. Distribute the base model + tooling, with which a user can specialize the model: The tooling takes the base model and trains it with code samples in the programming language relevant for a project, and documents and code from the problem domain. There could be pre-made collections that extend the license coverage (e.g. it should be ok to train with GPLv3 code when the model is used to develop GPLv3 code, I guess?) Everything that's put into it gets recorded in the CREDITS. When done, prune the model and quantize it down, so it ends up taking 50MB (for example), then test the result. If not good enough, train some more, repeat until satisfied. This process could take a week or three on regular hardware, but the result would be a model that runs locally, is tuned towards the problem domain, and is small. As such it should be able to support the developer who created it, while independent from Big Tech², without huge hardware investments³, and with proper attribute for everything that went into it (the CREDITS file, which would have to be amended during the specialization phase) And when working in different domains, you can just keep multiple such models around, then activate the right one as you switch projects - that way the models can remain small without losing too much precision. So, thoughts? ¹ I still have hopes for ~November for Reasons™, and the time line even seems to accelerate. ² In a way, this idea might be similar to how the home computer and personal computer revolution took power away from the mainframe crowd - even though mainframes and centralized systems exist and play a role in society to this day, a lot that happened in computering in the last ~40 years wouldn't have happened without the Amiga and the Am386 processor, respectively. ³ Limiting the gate-keeping effect of technology that requires unobtainium-grade GPU hardware to use.

NPR: Millions of people are pretending to be AI chatbots — for fun. “More than one third of U.S. adults have used ChatGPT, according to a June 2025 Pew Research study. People are not only deploying AI chatbots for everything from planning trips to doing homework assignments — they are also having fun impersonating them.” https://rbfirehose.com/2026/04/19/npr-millions-of-people-are-pretending-to-be-ai-chatbots-for-fun/

University of Michigan: A pocket-sized personal trainer: AI-written texts aim to get older adults moving. “Artificial intelligence can write text messages encouraging physical activity that most older adults consider appropriate and good quality, but their feelings about AI—and if they know AI wrote the message—impact their response, a new study in the Journals of Gerontology suggests.” https://rbfirehose.com/2026/04/17/a-pocket-sized-personal-trainer-ai-written-texts-aim-to-get-older-adults-moving-university-of-michigan/

Your Chatbot Isn’t a Therapist https://web.brid.gy/r/https://www.nytimes.com/2026/03/29/opinion/chatbot-therapy-ai.html

Is a random human peer better than a highly supportive chatbot in reducing loneliness over time? "The present study provides initial evidence that texting daily with a random human peer may be more effective in alleviating loneliness than texting with a highly supportive chatbot." Li, R.-N. et al. (2026) 'Is a random human peer better than a highly supportive chatbot in reducing loneliness over time?,' Journal of Experimental Social Psychology, 125, p. 104911. https://doi.org/10.1016/j.jesp.2026.104911. #OpenAccess #OA #Article #Psychology #AI #ArtificialIntelligence #Technology #Tech #Chatbots #Social #Loneliness #Academia

University of British Columbia: Texting with a stranger beats a chatbot at easing loneliness. “Texting with a real person—even a stranger—may reduce loneliness more than chatting with a highly supportive AI chatbot, new UBC research suggests.” https://rbfirehose.com/2026/04/07/university-of-british-columbia-texting-with-a-stranger-beats-a-chatbot-at-easing-loneliness/

🤖 Does an "AI messenger" exist? Curious if anyone has found anything like this in their journeys: Instead of sending a big long email or document to a colleague and having them not read it, what if you sent an agent of sorts inst... 📰 Source: Artificial Intelligence (AI) 🔗 Link: https://www.reddit.com/r/artificial/comments/1sozgok/does_an_ai_messenger_exist/ #AI #ArtificialIntelligence

I found a good use for corporate #ai #chatbots... #PromptPhreaking If you found a good one....post it up. #2600 #keepHOPEalive