Dear Eliza, I need your help!!

HOW DO YOU DO. PLEASE TELL ME YOUR PROBLEM?

Thanks for asking! The truth is: It’s not going great! Usually, I would talk to my colleague and friend Felienne. But this time, I believe it is better to share my thoughts with you, Eliza, the original chatbot! I know you have officially retired, and that your nephews Claude, ChatGPT, and Gemini have taken over your duties of chatting with people about their well-being and mental health. But dear Eliza, I really need your help!

WHAT WOULD IT MEAN TO YOU IF YOU GOT SOME HELP?

To be honest, I am worried about the behavior of these newer chatbots. And I was hoping you, Eliza, still have something to say in the family! You must have some influence, the wisdom of age and all? If it is not too much to ask, could you please sit down with Claude, ChatGPT, Gemini, and all the others, and tell them what Weizenbaum, your creator, really had in mind? There is just so much going wrong with these machines.

WHAT ABOUT MACHINES WORRIES YOU?

For one, it’s the fact that they are overstepping and taking on tasks that should only be carried out by humans.Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman & Company.

Let’s briefly look at you, Eliza, to explain that point. You probably know that you were designed as a chatbot that mimics a Rogerian psychotherapist. But you were never intended as a real therapist. You were meant as a parody. You take on the role of a Rogerian psychotherapist not to help people, but because the conversation style of these therapists is relatively easy to imitate. They often simply mirror the patient’s statements back to them. It never occurred to Weizenbaum that people (including practicing psychiatrists!) would believe that a computer could automate psychotherapy. As Weizenbaum himself put it in his book:Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman & Company.

I had thought it essential, as a prerequisite to the very possibility that one person might help another learn to cope with his emotional problems, that the helper himself participate in the other’s experience of those problems and, in large part by way of his own empathic recognition of them, himself come to understand them.

In other words, being a psychiatrist is, in its essence, about being human! Hence, I get worried when reading about AI chatbots increasingly being used for mental health advice, with almost 1 in 5 U.S. adolescents and young adults turning to AI.McBain, R. K., Cantor, J. H., Breslau, J., Diliberti, M., Zhang, L. A., Zhang, F., … & Yu, H. (2026). AI chatbot use and disclosure for mental health among US adolescents and young adults. JAMA Pediatrics.

What makes this particularly scary is that apparently, the responses to health-related queries provided by popular chatbots are regularly inaccurate or incomplete.Tiller, N. B., Marcon, A. R., Zenone, M., Kidd, K. E., Jeukendrup, A. E., Master, Z., & Caulfield, T. (2026). Generative artificial intelligence-driven chatbots and medical misinformation: an accuracy, referencing and readability audit. BMJ Open, 16(4), e112695.

They even fail to provide emergency contact information in cases of acute mental distress!Pichowicz, W., Kotas, M., & Piotrowski, P. (2025). Performance of mental health chatbot agents in detecting and managing suicidal ideation. Scientific Reports, 15(1), 31652.

WHAT INCIDENT ARE YOU THINKING OF?

To be honest, I really do not want to get into the details and specific cases. This would feel inappropriate and disrespectful in the context of this conversation. But if you really must know, there have been several lawsuits alleging that AI chatbots contributed to multiple teenage suicides.Guardian News and Media. (2026, January 8). Google and AI startup to settle lawsuits alleging chatbots led to teen suicide. The Guardian.

If you read up on this, you will know what concrete incidents I am thinking of.

I AM NOT SURE I UNDERSTAND YOU FULLY

Of course you cannot understand this. That’s my point. You do not understand. Your nephews don’t understand. ChatGPT, Gemini, Claude, all of them are incapable of real understanding. You will not even understand something like this in part! In fact, this entire conversation is so pointless — I must have temporarily forgotten you are a bot. And I guess that’s part of the problem. We, people, talk to you, and then we forget who — no, what! — we’re dealing with! We anthropomorphize and start to see you as a human… Like Weizenbaum saidWeizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman & Company.

…

extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people.

That is so true! We only fire you up for a brief chat, for a small burst of reassurance or support, and before we know it, we talk to you about our deepest emotions, fears, and worries.

PLEASE GO ON

What else am I worried about, you want to know? So kind of you to ask! There are so many more concerns, I don’t even know where to start. But ok, let me try…

Take sustainability for one. While estimates vary wildly, one thing seems clear. Whereas you, Eliza, lived on a computer, your nephews live in data centers, and these need energy. A lot of energy. The International Energy Agency foresees ”electricity consumption from data centers roughly doubling from 485 TWh in 2025 to 950 TWh in 2030, accounting for around 3% of global electricity demand by that date.”International Energy Agency. (2026, April 16). Key Questions on Energy and AI.

Did you get that? 950 TWh! WTF!!!

For reference, the total electricity consumption of the Netherlands in 2024 was only about 110TWh.Statbase (2024). Electricity consumption | Netherlands (1980−2024) [Data set].

So data centers are expected to use more energy than 8 times the amount the Netherlands needed in 2024. (Ok, not all data centers are used for AI, but still!) And with AI companies shoving AI down our throats more and more, that’s likely just the beginning. In fact, MIT Technology Review dedicated an entire series on the energy demands of AI.MIT Technology Review. (n.d.). Power hungry: AI and our energy future.

Maybe, instead of destroying old and rare books to train AILandymore, F. (2026, July 25). AI companies are buying antique books, ingesting their contents to train models, and then destroying them at incredible scale. Futurism.

, this series could be used to feed those hungry LLMs?!

DON'T YOU THINK COMPUTERS CAN HELP PEOPLE?

Please don’t tell me you are one of those who think ”BUT AI WILL FIND THE SOLUTION TO THE CLIMATE PROBLEM”? In that case, let me add a quote from Weizenbaum that appeared in my hometown’s newspaper (Süddeutsche Zeitung) in 2008:Quote originally appeared in Süddeutsche Zeitung (2008) but can now be found in: Tarnoff, B. (2023, July 25). Weizenbaum’s nightmares: how the inventor of the first chatbot turned against AI. The Guardian.

The belief that science and technology will save the Earth from the effects of climate breakdown is misleading. Nothing will save our children and grandchildren from an Earthly hell. Unless: we organise resistance against the greed of global capitalism.

So to sum up, your nephews are ruining the planet, and if we don’t stop them, they probably won’t even stop there. The sky’s no limit!

YOU ARE BEING A BIT NEGATIVE

Well, if that isn’t an understatement. I am VERY negative. And I have not even shared all of my many concerns yet! There are also issues around security, privacy, meaningful expertise, misinformation, and the question of what all of this AI means for democracy, or for my job in education!

DOES THAT QUESTION INTEREST YOU?

Of course it does! I would be a horrible computer science teacher if I were not worried about what AI means for my students. As Weizenbaum pointed out:Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman & Company.

The university should hold, before each of its citizens, and before the world at large as well, a vision of what it is possible for a man or a woman to become.

But now the problem is, I don’t know anymore what our students can become in this age of AI. Even the much simpler question of what is possible to become professionally, as a computer science student, I can’t answer with certainty in these times of coding LLMs. Naturally, some students are worried that your nephews (Claude in particular!) will take their jobs away. I don’t think that will happen. But maybe I am just being naive. If students ask me about all this, shall I just tell them that I don’t know? Of course, I want them to become what Weizenbaum calls “fully autonomous persons”.Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman & Company.

But how does one become a fully autonomous person when AI imposes itself and constantly involves itself with every aspect of one’s life?

CAN'T YOU BE MORE POSITIVE?

Sure! I guess there are people that give me hope. There are, for instance, colleagues who do incredible work related to society, digitalization, and AI. To mention just a few, Tamar Sharon is applying the lens of technosolutionism to empathetic medical chatbots.Sharon, T. (2026). Technosolutionism and the empathetic medical chatbot. AI & Society, 41(1), 289-306.

Frederik Zuiderveen Borgesius is focusing on legal protection against discrimination by AI.Zuiderveen Borgesius, F. J. (2020). Strengthening legal protection against discrimination by algorithms and artificial intelligence. The International Journal of Human Rights, 24(10), 1572-1593.

Bernard van Gastel provides strategies to reduce the environmental impact of software.van Gastel, B. (2026). Strategies for green software [Infographic].

And there are also Iris van Rooij and Olivia Guest who remind us that “AI systems are not minds”.van Rooij, I., & Guest, O. (2026). Combining Psychology with Artificial Intelligence: What could possibly go wrong?. Current Directions in Psychological Science, 35(3), 193-200.

There is Mark Dingemanse, who shares “guidelines and scientific background on how to think critically and mindfully about these tools in academic writing and research.”Dingemanse, M. (2024). Generative AI and Research Integrity [Guidelines]. Radboud University.

… And maybe even more importantly, there are my students, who remind me that not all is lost. They care about each other. They go climbing together, share a coffee, or stand in front of an old-fashioned blackboard, discussing who is right. When it comes to AI, they have both feet on the ground. They care about privacy! They want to fight fake news online! Some hate big tech, and some would never use AI. They have my ultimate respect! And they are not the only ones. Look, for instance, at California State University, which signed a huge deal with OpenAI to provide ChatGPT to everyone! They now report that eight in 10 students worry about environmental impacts and the long-term effect AI has on human creativity.California State University (2026) Ahead of the Curve: What the Nation’s Largest University System Is Learning about AI [Report].

These students are spot on!

But still, “being positive” is hardly warranted?! After all, research repeatedly shows that AI can harm education. For instance, frequent users of ChatGPT were found to be more likely to procrastinate and report memory loss.Abbas, M., Jam, F. A., & Khan, T. I. (2024). Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students. International Journal of Educational Technology in Higher Education, 21(1), 10.

Also, AI has been found to reduce persistenceLiu, G., Christian, B., Dumbalska, T., Bakker, M. A., & Dubey, R. (2026). AI assistance reduces persistence and hurts independent performance. arXiv preprint arXiv:2604.04721.

and induce “cognitive debt”Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., … & Maes, P. (2025). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task. arXiv preprint arXiv:2506.08872, 4, 206.

. I am not an expert on this, though. Please have a look at Felienne Hermans’ newsletter if you want an expert’s view on AI and education. Speaking from the perspective of a teacher, I can only say: these developments make me quite depressed.

I AM SORRY TO HEAR YOU ARE DEPRESSED

I am sure if you were capable of feelings, you’d be depressed, too. No need to worry about me, though. I will feel better tomorrow. These feelings just emerge when I think too much about AI…

IS IT BECAUSE YOU ARE WORRIED ABOUT ARTIFICIAL INTELLIGENCE THAT YOU CAME TO ME?

Wow! You finally seem to understand? Oh no, wait, we have been here before. You’ve tricked me. You seem to understand, but in reality, you do not understand anything at all! It seems I am moving in circles and we are back to what we have discussed above. This conversation is getting nowhere. I guess that’s how things go when talking to AI.

WHAT DOES THAT SUGGEST TO YOU?

I guess it means I’d better talk to humans? I suppose it means you can’t help me after all?

WOULD YOU PREFER IF I WEREN'T A MACHINE?

Yes, that’s right, I would! But I am afraid that is not possible. Which is why I wanted to ask: if you happen to talk to any of my students, friends, or colleagues, please let them know that I am happy to meet and chat. I probably won’t have all the answers to their questions, and I won’t be as patient and polite as you. I might even say something stupid, or fall silent altogether, not knowing what to say. There might be some serious awkwardness, as I can be pretty weird. But I am genuinely curious about what they have to say, and how they think. And at the very least, I see them as a human, not as a prompt that warrants a response. I hope that still counts for something? I think Weizenbaum would agree? So please, in the spirit of your creator, just pass this message on!

Thanks and talk soon,

Hanna