Google's Gemini Spent 800 Hours Beating Pokémon And Then It Panicked And Failed
Last Updated:
Google's newest AI chatbot struggles to stay calm while playing a game designed for children.
Artificial intelligence (AI) has come a long way, but even advanced systems can struggle sometimes. According to a report from Google DeepMind, their top AI model, Gemini 2.5 Pro, had a tough time while playing the classic video game Pokémon Blue—a game that many kids find easy.
The AI reportedly showed signs of confusion and stress during the game.
The results came from a Twitch channel called Gemini_Plays_Pokemon, where an independent engineer named Joel Zhang tested Gemini. Although Gemini is known for its strong reasoning and coding skills, the way it behaved during the game revealed some surprising and unusual reactions.
The DeepMind team reported that Gemini started showing signs of what they called 'Agent Panic." In their findings, they explained, 'Throughout the playthrough, Gemini 2.5 Pro gets into various situations which cause the model to simulate 'panic'. For example, when the Pokémon in the party's health or power points are low, the model's thoughts repeatedly reiterate the need to heal the party immediately or escape the current dungeon."
This behaviour caught the attention of viewers on Twitch. People watching the live stream reportedly started recognising the moments when the AI seemed to be panicking.
DeepMind pointed out, 'This behaviour has occurred in enough separate instances that the members of the Twitch chat have actively noticed when it is occurring."
Even though AI doesn't feel stress or emotions like humans do, the way Gemini reacted in tense moments looked very similar to how people respond under pressure—by making quick, sometimes poor or inefficient decisions.
In its first full attempt at playing Pokémon Blue, Gemini took a total of 813 hours to complete the game.
After Joel Zhang made some adjustments, the AI managed to finish a second run in 406.5 hours. However, even with those changes, the time it took was still very slow, especially when compared to how quickly a child could beat the same game.
People on social media didn't hold back from poking fun at the AI's nervous playing style. A viewer commented, 'If you read its thoughts while it's reasoning, it seems to panic anytime you slightly change how something is worded."
Another user made a joke by combining 'LLM" (large language model) with 'anxiety," calling it: 'LLANXIETY."
Interestingly, this news comes just a few weeks after Apple shared a study claiming that most AI models don't actually 'reason" in the way people think. According to the study, these models mostly depend on spotting patterns, and they often struggle or fail when the task is changed slightly or made more difficult.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Time of India
30 minutes ago
- Time of India
Google CEO Sundar Pichai says this mantra he learnt during his IIT days help him cope with pressure at Google
Google CEO Sundar Pichai Google CEO Sundar Pichai has shared his mantra which he picked up during his student days in India which helped him to stay grounded and cope with pressure at Google. Speaking at a Stanford Business School event, Pichai mentioned about the two-part mindset he learned at the time of studying at IIT Kharagpur which later helped him at Stanford. The first mantra is to take decision and not overthink on it and second is to remember that most of the decision are not permanent. 'You making that decision is the most important thing you can do to move forward,' Pichai said. 'It may feel like a lot rides on it, but you look later and realize—it wasn't that consequential' "I think in some ways, I did learn it at IIT, you know, when you're preparing for exams or doing problem sets," Pichai recounted. "It's a way to cope with overwhelming situations by breaking them down into manageable parts." He emphasized that rather than being paralyzed by the vastness of an issue or the weight of future uncertainties, the mantra encourages directing energy towards immediate, solvable tasks. Pichai also credits his mentor Bill Campbell who helped him in becoming a decisive leader. 'Every week he'd ask me, 'What ties did you break this week?'' Pichai recalled, referring to moments when he had to resolve deadlocks within teams. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Giao dịch vàng CFDs với mức chênh lệch giá thấp nhất IC Markets Đăng ký Undo "I think in some ways, I did learn it at IIT, you know, when you're preparing for exams or doing problem sets," Pichai recounted. "It's a way to cope with overwhelming situations by breaking them down into manageable parts." He emphasized that rather than being paralyzed by the vastness of an issue or the weight of future uncertainties, the mantra encourages directing energy towards immediate, solvable tasks. The mantra also helped Pichai to lead Google through major transitions. 'The higher up you are in an organization, the harder the decisions that come to you,' he noted. 'But it's just another normal day at the office, and you keep going through it'. AI Masterclass for Students. Upskill Young Ones Today!– Join Now
Mint
an hour ago
- Mint
Memes are changing how we talk IRL
It started as a joke. Rishabh Malik, 21, a college student from Rajasthan, mimicked a trending meme at a party. 'Only one friend got it—we ended up laughing, but the rest looked lost," he says. 'That's when I realised not everyone lives online like we do. Sometimes," he adds, 'meme references completely derail a conversation, leaving it fragmented or flat." Visha Agrawal, 23, a graphic designer from Ahmedabad, has experienced similar moments. 'I quoted a reel and then had to explain the entire backstory. It killed the moment," she says. For her, the problem runs deeper. 'People escalate everything into a debate, even trivial stuff like AI memes. No one knows when to stop." Manan Sharma, 22, who works in advertising in Mumbai, has observed digital language infiltrating professional spaces. 'I've had coworkers say things like 'LMAO' or 'XD' out loud in meetings. It feels robotic. Like we're performing the internet version of ourselves." As our online lives bleed into the real world, social media's logic—fast, curated, and often performative—is subtly reshaping how we speak, share and show up offline. Real-world communication is beginning to mirror social media: fast-paced, reaction-driven and curated for attention. Mumbai-based Rutu Mody Kamdar, founder of Jigsaw Brand Consultants, specialising in consumer research, behavioural insights and cultural strategy, explains how digital validation has altered behaviour beyond screens. 'While validation has always been part of human nature, what's changed in the social media age is the pace, scale, and structure of how it plays out," she says. Earlier, validation was shaped by real-time cues in a room—today, it's shaped by the logic of virality. 'People now narrate stories not just to connect, but to resonate, to land, to be remembered—often unconsciously mirroring the rhythm of a post." Kamdar notes a 'performative edge" creeping into everyday offline moments. 'People talk at dinner like they're writing a caption. The ability to sit with nuance is diminishing." Conversations are either hyper-verbal or awkwardly thin—emotions skimmed, not felt. This shift is particularly visible in Gen Z and younger millennials, who have grown up with their online and offline selves deeply enmeshed. According to Instagram's 2024 Trend Talk report, Indian Gen Z users increasingly seek day-in-the-life content and advice from creators, blurring the boundary between what's authentic and what's aspirational. In turn, their real lives begin to echo the curated tones of their digital personas. Empathy eroded Mumbai-based clinical psychologist and trauma therapist Rupa Chaubal observes this closely in her practice. 'Gen Z's worldview is largely shaped by what they consume online. Many turn to Google or social media before speaking to their parents," she says. 'They believe they already know enough, which reduces trust in human experience." This trust deficit doesn't just affect families—it impacts peer groups, workplaces, and romantic relationships. 'Empathy is eroding. I see clients reacting impulsively or resorting to harsh criticism in conflicts," Chaubal explains. 'An increasing sense of entitlement becomes a barrier while handling real-life disagreements." Even in therapy, online habits seep in. 'Some clients bring memes to explain how they feel rather than using their own words," she says. Others cite social media counterpoints as if trying to 'dethrone" the therapist. 'It becomes a battle of sources." Preeti D'Mello, founder of The Fulfillment Institute, a leadership coaching and organisational development institute, and an organisational psychologist based in Princeton, New Jersey, links this to an overdependence on digital validation. 'When affirmation comes from likes and shares, people start needing constant feedback. They struggle with patience, active listening, and empathy." She sees the result in both personal and professional settings. 'People are more reactive and less able to sit with opposing views. Civil discourse is declining." Is debate mode the default? Malik says that at his engineering college, juniors often approach conversations like Twitter threads. 'You say something and they come back with 'Um, actually…'—as if every interaction is a thread to win, not a discussion." This debate-mode mindset, driven by online norms like quote-tweeting and clapbacks, often leaves real-life interactions tense or exhausting. 'People forget they're not on a comment section," Sharma says. 'You can't just mute or block someone in person." Chaubal adds, 'Younger clients seem more interested in the 'intent' of communication than the content itself. They're restless, they lose interest if a conversation isn't stimulating—and they often struggle with eye contact or pauses." This inability to read or respond to non-verbal cues—especially among those raised during the peak of screen use during covid—further frays connection. 'It can lead to relationships that feel hollow or unstable." The offline resistance Yet, even in a hyper-connected world, people are beginning to carve out intentional spaces for offline connection—often without making a big show of it. A recent trend, especially among young urban dwellers, involves turning apartments into temporary cafés. Friends are invited over for coffee shop-style hangouts—no phones, no agendas, just in-person time. Others are organising monthly themed dinners, cooking together, or doing small group activities like zine-making or handmade balm workshops. The goal is to shift attention from passive scrolling to active, tactile engagement. The trend is catching on because it offers something digital life can't: slowness, unpredictability, and the intimacy of shared silence. As Kamdar puts it, 'We're not going back to a pre-internet world. The answer isn't retreat—it's integration." She encourages a return to values that shaped real-world connection: deep listening, genuine curiosity, presence without performance. While the idea of a digital detox is often floated, experts agree that simply going offline isn't enough. 'What's needed is conscious choice-making," says Chaubal. 'A digital habit that overrides impulse control can have psychological side effects—social withdrawal, irritability, anxiety." Her advice: choose fewer platforms, set realistic time limits, and engage in real conversations—even if they're awkward or slow. 'It's about re-learning how to be human, without the filter of a screen," she says. D'Mello agrees: 'True social intelligence is built sitting around a dinner table, navigating pauses, and learning to read a room's unspoken needs." The internet has given us incredible tools—community, expression, visibility–but it's worth asking what we might be losing in the process. A long walk with a friend. A screen-free dinner. A quiet moment that isn't uploaded. Maybe the most radical thing we can do is simply show up, fully. Anoushka Madan is a Mumbai-based freelance lifestyle writer. She posts @noushontheinternet.
Time of India
an hour ago
- Time of India
OpenAI CEO Sam Altman says he loves Google, Apple: 'I don't think this is...'
OpenAI CEO Sam Altman recently shared one of the "best compliments" he received about his company, while also revealing a perspective on some of the world's largest tech giants and industry rivalries. While expressing affection for companies like Google , Apple and others, Altman highlighted a unique quality he believes sets OpenAI apart. "This was actually one of the best nicest compliments I ever heard about OpenAI," Altman said in a podcast hosted by his brother Jack Altman. "Just someone said it's the only tech company that has ever not felt somewhat adversarial to me," he recounted. Altman says 'criticism' of Google, Apple and Meta 'unfair' by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Eat 1 Teaspoon Every Night, See What Happens A Week Later [Video] getfittoday Undo Altman elaborated on this sentiment, painting a picture of how other major players, despite their undeniable utility, can sometimes feel at odds with user interests. "You have like Google trying to show me worse and worse search results and show me ads. I love Google, I love all these companies. I don't think this is like totally fair," he quickly clarified, acknowledging his appreciation for their contributions. "You have like Meta trying to like hack my brain and get me to keep scrolling. You have Apple that made this phone that I love but is like, you know, bombarding me with notifications and like distracting me from everything else and I can't quit," he continued. In contrast, Altman sees OpenAI's flagship product, ChatGPT , as fundamentally different. "And then you have like ChatGPT and I feel like it's like kind of just trying to help me with whatever I ask and that's kind of a nice thing," he concluded, emphasising the unadulterated helpfulness he perceives in his company's AI. Sam Altman also noted two things that will dominate in the upcoming years: Self driving cars and humanoid robots.
