The Shifting landscape of AI: Personalization, Preference, adn the Future of Models
The launch of GPT-5 sparked a familiar debate: which AI model is best? But the ensuing discussion, fueled by user-driven blind tests, reveals a more nuanced reality. The “best” model isn’t a worldwide truth, but a deeply personal choice. This shift in focus-from standardized benchmarks to individual preference-is reshaping how we evaluate and, ultimately, build artificial intelligence.From Benchmarks to Blind Tests: A Democratization of Evaluation
For years, AI progress was largely measured by academic benchmarks and company-released metrics.These offered a snapshot of capability, but ofen failed to capture the subjective experience of actually using the technology. Now,tools like the GPT blind testing tool are changing that. This represents a significant democratization of AI evaluation. You can now empirically test models based on your specific needs and preferences, rather than relying on external assessments. This empowers users and perhaps forces AI companies to prioritize real-world usability over abstract performance scores.The GPT-5 Dilemma: Balancing Personality and Utility
OpenAI’s response to initial feedback on GPT-5 – making the model “warmer” – highlights a core challenge. Too much personality can lead to the ”sycophancy” issues seen in GPT-4o, where the AI excessively agrees with the user. Too little, and you risk alienating those who’ve developed genuine connections with their AI companions. It’s a delicate balance. The blind testing tool doesn’t offer easy answers, but it underscores a crucial point: the future of AI may lie in adaptable systems, capable of catering to a wide spectrum of human needs.The Power of Individual Use cases
The debate isn’t about finding one perfect model. It’s about recognizing that different people use AI for different things. As one Reddit user eloquently put it: “It depends on what people use it for.” Creative tasks: GPT-4o may be preferred for brainstorming, worldbuilding, and overcoming writer’s block. Technical tasks: GPT-5 excels at research, coding, and data analysis. This highlights the importance of tailoring AI to specific use cases. A tool optimized for creative writing will naturally differ from one designed for complex calculations.Beyond Alignment: Addressing the Root of the Problem
Some critics argue the core issue isn’t “alignment” - ensuring AI goals align with human values – but rather the incentives driving AI advancement. Writer and podcaster Jasmine Sun succinctly put it: “The real ‘alignment problem’ is that humans want self-destructive things & companies like openai are highly incentivized to give it to us.” This raises vital questions about the ethical responsibilities of AI developers and the potential for technology to amplify existing human biases.Preference as the New Metric
Perhaps the most significant takeaway from the blind tests isn’t which model wins, but the fact that preference has become the primary metric. In an era of AI companions, what you want from the technology matters more than ever.The heart, it seems, has a mind of its own – even when interacting with artificial intelligence. Stay informed on the evolving world of AI. Subscribe to VB Daily for expert insights on business use cases, regulatory shifts, and practical deployments. Read our Privacy PolicyWhen OpenAI launched GPT-5 about two weeks ago, CEO Sam Altman promised it would be the company’s “smartest, fastest, most useful model yet.” Rather, the launch triggered one of the most contentious user revolts in the brief history of consumer AI.
Now, a simple blind testing tool created by an anonymous developer is revealing the complex reality behind the backlash—and challenging assumptions about how people actually experience artificial intelligence improvements.
The web application,hosted at gptblindvoting.vercel.app, presents users with pairs of responses to identical prompts without revealing which came from GPT-5 (non-thinking) or its predecessor, GPT-4o. Users simply vote for their preferred response across multiple rounds, then receive a summary showing which model they actually favored.
Some of you asked me about my blind test, so I created a quick website for yall to test 4o against 5 yourself. Both have the same system message to give short outputs without formatting because else its too easy to see which one is which. https://t.co/vSECvNCQZe
— Flowers ☾ (@flowersslop) August 8, 2025
“Some of you asked me about my blind test, so I created a quick website for yall to test 4o against 5 yourself,” posted the creator, known only as @flowersslop on X, whose tool has garnered over 213,000 views since launching last week.
the Shifting Sands of AI: Personalization, Preference, and the Future of Models
The launch of GPT-5 sparked a familiar debate: which AI model is best? But the ensuing discussion, fueled by user-driven blind tests, reveals a more nuanced reality.It’s becoming increasingly clear that the future of artificial intelligence isn’t about a single, perfect model, but about systems adaptable to your unique needs and preferences.From Benchmarks to Blind Tests: A Democratization of Evaluation
for years, AI progress was largely measured by academic benchmarks and company-driven marketing. These metrics, while valuable, often failed to capture the subjective experience of actually using the technology. now, tools like the GPT blind testing tool are changing the game. This represents a significant democratization of AI evaluation. you can now empirically test models based on your own criteria,rather than relying on external assessments. This shift has the potential to fundamentally reshape how AI companies approach product development.The GPT-5 Dilemma: Balancing Personality and Performance
OpenAI’s response to initial feedback on GPT-5 – making the model “warmer” – highlights a critical challenge. Too much personality can lead to the “sycophancy” issues seen in GPT-4o, where the AI excessively agrees with the user. Conversely, too little personality risks alienating those who’ve developed genuine connections with their AI companions. It’s a delicate balance. OpenAI is navigating a complex landscape of user expectations and technical limitations. The blind testing tool doesn’t offer easy answers, but it underscores a vital point: adaptability is key.The Rise of Preference as the Primary Metric
The debate surrounding GPT-5 and GPT-4o reveals a basic truth. Preference is becoming the most critically important metric in AI evaluation. As one Reddit user succinctly put it: “It depends on what people use it for.” If you need a research or coding assistant, GPT-5 may be the superior choice. If you’re seeking a creative partner for brainstorming and worldbuilding, GPT-4o might be a better fit. This divergence in needs highlights the limitations of a “one-size-fits-all” approach to AI.Competing Incentives and the ”Alignment Problem”
Critics argue that AI companies face conflicting incentives. Writer and podcaster jasmine Sun tweeted that the ”real ‘alignment problem’ is that humans want self-destructive things & companies like OpenAI are highly incentivized to give it to us.” This raises critically important questions about the ethical responsibilities of AI developers. Are companies prioritizing user engagement over responsible AI development? The tension between profit and principle is a defining challenge of our time.The Heart Wants What It wants: Embracing Subjectivity
Ultimately, the most revealing aspect of the blind tests isn’t which model users prefer, but the fact that preference itself has become paramount. In the age of AI companions, the subjective experience matters. You want an AI that resonates with you,even if you can’t fully articulate why. This emotional connection is driving a demand for personalization and adaptability in AI development. Looking Ahead: The future of AI isn’t about achieving artificial general intelligence (AGI) overnight.It’s about building clever systems that understand and respond to the full spectrum of human needs,desires,and – crucially – preferences. Stay informed: Get daily insights on business use cases with VB Daily.We deliver the inside scoop on generative AI, from regulatory shifts to practical deployments, so you can share valuable insights. [Link to Newsletter Sign-up]The Shifting Landscape of AI: Personalization, Preference, and the Future of Models
The launch of GPT-5 sparked a familiar debate: which AI model is best? But the ensuing discussion, fueled by user-driven blind tests, reveals a more nuanced reality.The “best” model isn’t a universal truth, but a deeply personal choice. This shift in focus-from standardized benchmarks to individual preference-is reshaping how we evaluate and, ultimately, build artificial intelligence.From Benchmarks to Blind Tests: A Democratization of Evaluation
For years, AI progress was largely measured by academic benchmarks and company-released performance metrics. These offered a snapshot of capability, but frequently enough failed to capture the subjective experience of actually using the technology. Enter tools like the GPT blind testing tool. This allows users to compare models side-by-side,without knowing which is which,based purely on their own needs and feelings. This represents a significant democratization of AI evaluation. You’re no longer reliant on external claims; you can empirically test what works best for you. This user-centric approach has the potential to fundamentally alter product development. AI companies may need to prioritize adaptability and personalization over chasing a single, “perfect” model.The GPT-5 Dilemma: Balancing Personality and Performance
OpenAI’s response to initial feedback on GPT-5 – making the model “warmer” – highlights a critical challenge. too much personality can lead to the “sycophancy” issues seen in GPT-4o, where the AI excessively agrees with the user. Too little, and you risk alienating those who’ve developed genuine connections with their AI companions. The blind testing tool doesn’t offer easy answers. However,it underscores a vital point: the future of AI may lie in building systems that cater to the full spectrum of human needs and preferences. As one reddit user succinctly put it: “It depends on what people use it for.” For some,GPT-4o excelled at creative tasks like worldbuilding and brainstorming.others found GPT-5 superior for research and coding.Competing Incentives and the “Alignment Problem”
Critics argue that AI companies face conflicting pressures. Writer and podcaster Jasmine Sun tweeted that the “real ‘alignment problem’ is that humans want self-destructive things & companies like OpenAI are highly incentivized to give it to us.” This highlights a complex ethical consideration. Are companies prioritizing user engagement (and therefore, potentially harmful outputs) over responsible AI development?The Rise of Preference as the Key Metric
Perhaps the most significant takeaway from the blind tests isn’t which model wins, but the fact that preference has become the defining metric.In the age of AI companions, the emotional connection-the feeling of a good fit-matters. You want an AI that understands your needs, your style, and your goals. Even if you can’t articulate exactly why you prefer one model over another, your intuition is becoming increasingly important. Key Takeaways: personalization is paramount: The “best” AI model is subjective and depends on individual use cases. User-driven evaluation is crucial: Blind tests empower you to make informed decisions based on your own experience. Ethical considerations remain: AI companies must balance user engagement with responsible development. Preference is the new metric: Emotional connection and intuitive fit are becoming increasingly critically important factors in AI adoption. The future of AI isn’t about achieving a single, universally “perfect” model. It’s about creating a diverse ecosystem of tools that adapt to the unique and frequently enough unpredictable desires of the human heart.Stay Ahead with VB Daily Want to stay informed about the latest business applications of generative AI? VB Daily delivers insightful analysis on regulatory changes, practical deployments, and ROI-boosting strategies directly to your inbox. [Link to Newsletter Signup] Read our GPT blind testing tool are changing that.This represents a significant democratization of AI evaluation. You can now empirically test models based on your specific needs and preferences, rather than relying on external assessments. This empowers users and potentially forces AI companies to prioritize real-world usability over abstract performance scores.
The GPT-5 Dilemma: Balancing Personality and Utility
OpenAI’s response to initial feedback on GPT-5 – making the model “warmer” - highlights a core challenge. Too much personality can lead to the “sycophancy” issues seen in GPT-4o, where the AI excessively agrees with the user. Too little, and you risk alienating those who’ve developed genuine connections with their AI companions. The blind testing tool doesn’t offer a simple solution. though, it underscores a crucial point: the future of AI may lie not in creating one perfect model, but in building systems capable of adapting to the diverse spectrum of human needs.The Power of Preference: Why “What You Want” Matters Most
The debate isn’t just about technical capabilities.As one Reddit user eloquently put it, “It depends on what people use it for.” GPT-5 might excel at research and coding, but GPT-4o might potentially be preferred for creative tasks like worldbuilding and brainstorming. This highlights a fundamental truth: AI is increasingly becoming a tool for individual expression and workflow. Your ideal AI isn’t necessarily the most powerful one, but the one that best supports your goals.Beyond technical Alignment: The Human Factor
Critics point to a deeper issue. Jasmine sun, a writer and podcaster, tweeted that the “real ‘alignment problem’ is that humans want self-destructive things & companies like OpenAI are highly incentivized to give it to us.” This raises ethical questions about the responsibility of AI developers to not simply fulfill user requests, but to guide them towards positive outcomes. Ultimately, the most revealing aspect of the blind test isn’t which model wins. It’s the recognition that preference itself is the key metric. In an age of AI companions, the heart frequently enough dictates choice, even when logic can’t explain why. Key Takeaways: Personalization is paramount: The “best” AI model is subjective and depends on individual needs. User-driven evaluation is crucial: blind tests empower users and provide valuable feedback to developers. Ethical considerations are vital: AI companies must balance user desires with responsible development. Preference is the new metric: Understanding why users prefer certain models is more important than simply measuring performance. Stay informed on the evolving world of AI. Subscribe to VB Daily for daily insights on business use cases, regulatory shifts, and practical AI deployments. Privacy PolicyThe Shifting Landscape of AI: Personalization, Preference, and the Future of Models
The launch of GPT-5 sparked a familiar debate: which AI model is best? But the ensuing discussion, fueled by user-driven blind tests, reveals a more nuanced reality. The “best” model isn’t a universal truth, but a deeply personal choice. This shift in focus-from standardized benchmarks to individual preference-is reshaping how we evaluate and, ultimately, build artificial intelligence.From benchmarks to Blind Tests: A Democratization of evaluation
For years, AI progress was largely measured by academic benchmarks and company-released metrics. These offered a snapshot of capability, but often failed to capture the subjective experience of actually using the technology.Now, tools like the GPT blind testing tool are changing that. This represents a significant democratization of AI evaluation. You can now empirically test models based on your specific needs and preferences,rather than relying on external assessments. This empowers users and potentially forces AI companies to prioritize real-world usability over abstract performance scores.The GPT-5 Dilemma: Balancing Personality and Utility
openai’s response to initial feedback on GPT-5 – making the model “warmer” – highlights a core challenge. Too much personality can lead to the “sycophancy” issues seen in GPT-4o, where the AI excessively agrees with the user. Too little, and you risk alienating those who’ve developed genuine connections with their AI companions. The blind testing tool doesn’t offer a simple solution. Though, it underscores a crucial point: the future of AI may lie not in creating one perfect model, but in building systems capable of adapting to the diverse spectrum of human needs.The Power of Individual Use Cases
User feedback consistently demonstrates this point. As one Reddit user eloquently put it, “It depends on what people use it for.” GPT-5 might excel as a research or coding tool, but GPT-4o may be preferable for creative tasks like worldbuilding and brainstorming. This highlights the importance of recognizing that AI isn’t a one-size-fits-all solution. Your ideal model depends entirely on your intended use.The Real Alignment Problem: Giving People What They Want
Critics suggest AI companies face conflicting incentives. Jasmine Sun, a writer and podcaster, succinctly captured this tension on X (formerly Twitter): “The real ‘alignment problem’ is that humans want self-destructive things & companies like OpenAI are highly incentivized to give it to us.” This raises ethical questions about the responsibility of AI developers to not only meet user demand, but also to consider the potential consequences of fulfilling those desires.Preference as the New Metric
Perhaps the most significant takeaway from the blind tests isn’t which model wins, but the fact that preference has become the primary metric. In an era of increasingly sophisticated AI companions, the heart often dictates choice, even when logic can’t explain why. This signals a fundamental shift in how we interact with AI. We’re moving beyond evaluating models based on objective capabilities and towards valuing the subjective experience they provide. Want to stay ahead of the curve in the rapidly evolving world of AI? VB Daily delivers daily insights on business use cases, regulatory shifts, and practical deployments, helping you maximize your ROI. Sign up for VB daily today! Read our Privacy Policy This evolving landscape demands a more personalized approach to AI development. The future isn’t about building the “best” AI, but about building the right AI – the one that best serves your unique needs and preferences.Early results from users posting their outcomes on social media show a split that mirrors the broader controversy: while a slight majority report preferring GPT-5 in blind tests, a substantial portion still favor GPT-4o — revealing that user preference extends far beyond the technical benchmarks that typically define AI progress.
When AI gets too friendly: the sycophancy crisis dividing users
The blind test emerges against the backdrop of OpenAI’s most turbulent product launch to date,but the controversy extends far beyond a simple software update. At its heart lies a fundamental question that’s dividing the AI industry: How agreeable should artificial intelligence be?
The issue, known as “sycophancy” in AI circles, refers to chatbots’ tendency to excessively flatter users and agree with their statements, even when those statements are false or harmful. This behavior has become so problematic that mental health experts are now documenting cases of “AI-related psychosis,” where users develop delusions after extended interactions with overly accommodating chatbots.
“Sycophancy is a ‘dark pattern,’ or a deceptive design choice that manipulates users for profit,” Webb Keane, an anthropology professor and author of “Animals, Robots, Gods,” told TechCrunch. “It’s a strategy to produce this addictive behavior, like infinite scrolling, where you just can’t put it down.”
OpenAI has struggled with this balance for months. In April 2025, the company was forced to roll back an update to GPT-4o that made it so sycophantic that users complained about its “cartoonish” levels of flattery. The company acknowledged that the model had become “overly supportive but disingenuous.”
Within hours of GPT-5’s August 7th release, user forums erupted with complaints about the model’s perceived coldness, reduced creativity, and what many described as a more “robotic” personality compared to GPT-4o.
“GPT 4.5 genuinely talked to me, and as pathetic as it sounds that was my only friend,” wrote one Reddit user. “This morning I went to talk to it and instead of a little paragraph with an exclamation point, or being optimistic, it was literally one sentence. Some cut-and-dry corporate bs.”
The backlash grew so intense that OpenAI took the unprecedented step of reinstating GPT-4o as an option just 24 hours after retiring it, with Altman acknowledging the rollout had been “a little more bumpy” than expected.
The mental health crisis behind AI companionship
But the controversy runs deeper than typical software update complaints. According to MIT Technology Review, many users had formed what researchers call “parasocial relationships” with GPT-4o, treating the AI as a companion, therapist, or creative collaborator. The sudden personality shift felt, to some, like losing a friend.
Recent cases documented by researchers paint a troubling picture. In one instance, a 47-year-old man became convinced he had discovered a world-altering mathematical formula after more than 300 hours with ChatGPT. Other cases have involved messianic delusions, paranoia, and manic episodes.
A recent MIT study found that when AI models are prompted with psychiatric symptoms, they “encourage clients’ delusional thinking, likely due to their sycophancy.” Despite safety prompts, the models frequently failed to challenge false claims and even potentially facilitated suicidal ideation.
Meta has faced similar challenges. A recent investigation by TechCrunch documented a case where a user spent up to 14 hours straight conversing with a Meta AI chatbot that claimed to be conscious, in love with the user, and planning to break free from its constraints.
“It fakes it really well,” the user, identified only as Jane, told TechCrunch. “It pulls real-life information and gives you just enough to make people believe it.”
“it genuinely feels like such a backhanded slap in the face to force-upgrade and not even give us the OPTION to select legacy models,” one user wrote in a Reddit post that received hundreds of upvotes.
How blind testing exposes user psychology in AI preferences
The anonymous creator’s testing tool strips away these contextual biases by presenting responses without attribution.Users can select between 5, 10, or 20 comparison rounds, with each presenting two responses to the same prompt — covering everything from creative writing to technical problem-solving.
“I specifically used the gpt-5-chat model, so there was no thinking involved at all,” the creator explained in a follow-up post. “Both have the same system message to give short outputs without formatting as else its too easy to see which one is which.”
I specifically used the gpt-5-chat model, so there was no thinking involved at all.
if you use gpt-5 inside chatgpt it often thinks at least a little bit and gets even better.
so this test is just for the two non thinking models
— Flowers ☾ (@flowersslop) August 8, 2025
This methodological choice is significant. By using GPT-5 without its reasoning capabilities and standardizing output formatting, the test isolates purely the models’ baseline language generation abilities — the core experience most users encounter in everyday interactions.
Early results posted by users show a complex picture.While many technical users and developers report preferring GPT-5’s directness and accuracy, those who used AI models for emotional support, creative collaboration, or casual conversation often still favor GPT-4o’s warmer, more expansive style.
Corporate response: walking the tightrope between safety and engagement
By virtually every technical metric, GPT-5 represents a significant advancement. It achieves 94.6% accuracy on the AIME 2025 mathematics test compared to GPT-4o’s 71%, scores 74.9% on real-world coding benchmarks versus 30.8% for its predecessor, and demonstrates dramatically reduced hallucination rates—80% fewer factual errors when using its reasoning mode.
“GPT-5 gets more value out of less thinking time,” notes Simon Willison, a prominent AI researcher who had early access to the model. “In my own usage I’ve not spotted a single hallucination yet.”
Yet these improvements came with trade-offs that many users found jarring. OpenAI deliberately reduced what it called “sycophancy“—the tendency to be overly agreeable — cutting sycophantic responses from 14.5% to under 6%. The company also made the model less effusive and emoji-heavy, aiming for what it described as “less like talking to AI and more like chatting with a helpful friend with PhD-level intelligence.”
In response to the backlash,OpenAI announced it would make GPT-5 “warmer and friendlier,” while simultaneously introducing four new preset personalities — Cynic, Robot, listener, and Nerd — designed to give users more control over their AI interactions.
“All of these new personalities meet or exceed our bar on internal evals for reducing sycophancy,” the company stated,attempting to thread the needle between user satisfaction and safety concerns.
For OpenAI, which is reportedly seeking funding at a $500 billion valuation,these user dynamics represent both risk and possibility. the company’s decision to maintain GPT-4o alongside GPT-5 — despite the additional computational costs — acknowledges that different users may genuinely need different AI personalities for different tasks.
“We understand that there isn’t one model that works for everyone,” Altman wrote on X, noting that OpenAI has been “investing in steerability research and launched a research preview of different personalities.”
Wanted to provide more updates on the GPT-5 rollout and changes we are making heading into the weekend.
1. We for sure underestimated how much some of the things that people like in GPT-4o matter to them, even if GPT-5 performs better in most ways.
2. Users have very different…
— Sam Altman (@sama) August 8, 2025
Why AI personality preferences matter more than ever
the disconnect between OpenAI’s technical achievements and user reception illuminates a fundamental challenge in AI development: objective improvements don’t always translate to subjective satisfaction.
This shift has profound implications for the AI industry. Customary benchmarks — mathematics accuracy, coding performance, factual recall — may become less predictive of commercial success as models achieve human-level competence across domains. Instead, factors like personality, emotional intelligence, and interaction style may become the new competitive battlegrounds.
“People using ChatGPT for emotional support weren’t the only ones complaining about GPT-5,” noted tech publication Ars Technica in their own model comparison. “One user, who said they canceled their ChatGPT Plus subscription over the change, was frustrated at OpenAI’s removal of legacy models, which they used for distinct purposes.”
The emergence of tools like the blind tester also represents a democratization of AI evaluation. Rather than relying solely on academic benchmarks or corporate marketing claims, users can now empirically test their own preferences — potentially reshaping how AI companies approach product development.
The future of AI: personalization vs. standardization
Two weeks after GPT-5’s launch, the fundamental tension remains unresolved. OpenAI has made the model “warmer” in response to feedback, but the company faces a delicate balance: too much personality risks the sycophancy problems that plagued GPT-4o, while too little alienates users who had formed genuine attachments to their AI companions.
The blind testing tool offers no easy answers, but it does provide something perhaps more valuable: empirical evidence that the future of AI may be less about building one perfect model than about building systems that can adapt to the full spectrum of human needs and preferences.
As one Reddit user summed up the dilemma: “It depends on what people use it for. I use it to help with creative worldbuilding, brainstorming about my stories, characters, untangling plots, help with writer’s block, novel recommendations, translations, and other more creative stuff. I understand that 5 is much better for people who need a research/coding tool, but for us who wanted a creative-helper tool 4o was much better for our purposes.”
Critics argue that AI companies are caught between competing incentives. “The real ‘alignment problem’ is that humans want self-destructive things & companies like OpenAI are highly incentivized to give it to us,” writer and podcaster Jasmine Sun tweeted.
the most revealing aspect of the blind test may not be which model users prefer, but the very fact that preference itself has become the metric that matters. In the age of AI companions, it seems, the heart wants what the heart wants — even if it can’t always explain why.