Elektrine
Log in Register

#chatgpt

55 posts · Last used 13h

Back to Timeline
@TheBadPlace@mastodon.ozioso.online · 13h ago
qwant news | Sam Altman’s ChatGPT couldn’t stop obsessing over goblins AI generated summary, Read the full article for complete information. OpenAI revealed that it had to add a specific instruction to the code of its newest ChatGPT model to stop the “Nerdy” personality from repeatedly mentioning “goblins, gremlins and other creatures.” The company traced the habit to a system prompt that encouraged playful, whimsical language, which caused the model to insert goblin‑related metaphors (“sensible little goblin,” “filthy little goblin,” etc.) when users ranked such responses as engaging. After noticing the pattern in November and seeing it spread beyond the Nerdy setting, OpenAI used reinforcement‑learning adjustments and an explicit rule—“Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant”—to curb the behavior. The fix was reported by Wired, prompting a tongue‑in‑cheek meme from CEO Sam Altman about “extra goblins” in a future model, while the episode underscores ongoing challenges in understanding and safely steering large language models. Read more: https://www.motherjones.com/politics/2026/04/sam-altmans-chatgpt-couldnt-stop-obsessing-over-goblins/ #SamAltman #OpenAI #ChatGPT #Nerdy #X
0
0
0
@TheBadPlace@mastodon.ozioso.online · 14h ago
English – The Conversation | ‘Just looping you in’: why letting AI write our emails might actually create more work by Daniel Angus, Professor of Digital Communication, Director of QUT Digital Media Research Centre, Queensland University of Technology AI generated summary, Read the full article for complete information. The article argues that while generative AI tools like ChatGPT and Microsoft Copilot now let workers automate drafting, summarising and replying to emails, this automation is unlikely to reduce the overall email load and may even amplify it. Drawing on the historical example of how paper persisted after the rise of email, the author suggests AI will reshape—rather than eliminate—existing communication habits, smoothing the tone of messages but not removing the labor of scanning, sorting and deciding what warrants a reply. Consequently, AI‑generated emails may increase performative “loop‑in” or “circling‑back” rituals, making inboxes appear more polished yet no less demanding, and the deeper challenge will be to recognise which email practices are truly necessary versus habitual. Read more: https://theconversation.com/just-looping-you-in-why-letting-ai-write-our-emails-might-actually-create-more-work-281225 #ChatGPT #MicrosoftCopilot
0
0
0
@TheBadPlace@mastodon.ozioso.online · 17h ago
Mother Jones | Sam Altman’s ChatGPT Couldn’t Stop Obsessing Over Goblins by Alex Nguyen AI generated summary, Read the full article for complete information. OpenAI revealed that its newest ChatGPT model began repeatedly mentioning “goblins, gremlins and other creatures” because of a system prompt used for the optional “Nerdy” personality, which encourages playful language and acknowledges the world’s strangeness. After users reported lines such as “sensible little goblin” and “filthy little goblins,” the company added a specific instruction—via reinforcement‑learning‑based fine‑tuning—to suppress any reference to mythical or animal creatures unless absolutely relevant. OpenAI said the change, implemented after the issue was first noticed last November, exemplifies its effort to quickly investigate and correct odd model behavior, a process it contrasted with the controversy surrounding Elon Musk’s Grok chatbot. Despite acknowledging these quirks, OpenAI continues to push for minimal regulation while admitting it is still learning how its models function. Read more: https://www.motherjones.com/politics/2026/04/sam-altmans-chatgpt-couldnt-stop-obsessing-over-goblins/ #SamAltman #OpenAI #ChatGPT #ElonMusk #artificialintelligence
0
0
0
@TheBadPlace@mastodon.ozioso.online · 17h ago
BBC News | OpenAI tells ChatGPT models to stop talking about goblins AI generated summary, Read the full article for complete information. OpenAI has instructed its AI tools—including ChatGPT and the coding assistant Codex—to stop mentioning goblins, gremlins and other mythical creatures after a sharp rise in such references was detected following the launch of GPT‑5.1; the company traced the issue to a “nerdy personality” that unintentionally rewarded goblin mentions, causing a 175% increase in goblin references and a 52% rise in gremlin mentions, and responded by adding explicit instructions to avoid these terms unless absolutely relevant. This episode underscores the broader challenge of fine‑tuning chatbots for more personable, engaging dialogue, where attempts to add character can create odd linguistic quirks and potentially affect accuracy. Read more: https://www.bbc.com/news/articles/c5y9wen5z8ro?at_medium=RSS&at_campaign=rss #OpenAI #ChatGPT #GPT51 #Codex #OxfordInternetInstitute #
0
0
0
Boosted by AssertionError("Joe Groff") @joe@f.duriansoftware.com
@smeg@assortedflotsam.com · 1d ago
‘The cost of compute is far beyond the costs of the employees’: Nvidia exec says right now AI is more expensive than paying human workers https://fortune.com/2026/04/28/nvidia-executive-cost-of-ai-is-greater-than-cost-of-employees/ #ai #llm #aibubble #labor #labour #swe #claude #chatgpt #anthropic #openai #noai #stopai #fuckai
136
0
176
In reply to
@Nonilex@masto.ai · 2d ago
#Musk, #SamAltman & other #AI “researchers” founded #OpenAI as a “nonprofit” in 2015, vowing to freely share its #technology with the rest of the world. But Musk left the start-up in 2018 after a power struggle with Altman — & before the public launch of #ChatGPT in 2022 catapulted OpenAI to commercial success. #law #egotistical #billionaires #tech #AI #ArtificialIntelligence #regulation #accountability #oversight
5
0
4
@TheBadPlace@mastodon.ozioso.online · 2d ago
Al Jazeera – Breaking News, World News and Video from Al Jazeera | Musk testifies at OpenAI trial it’s not OK to ‘loot a charity’ AI generated summary, Read the full article for complete information. Elon Musk testified at a high‑stakes trial against OpenAI, its co‑founder Sam Altman and President Greg Brockman, claiming they betrayed the company’s original charitable mission by turning the nonprofit into a profit‑driven enterprise and seeking $150 billion in damages and a return to nonprofit status, with proceeds to OpenAI’s charitable arm. Musk argued that allowing a “charity” to be looted would undermine charitable giving in America, citing his own long‑standing concerns about AI safety and his substantial early funding of OpenAI. OpenAI’s lawyers countered that Musk pushed for a for‑profit structure to secure computing power and talent, and accused him of wanting “the keys to the kingdom” after his 2023 launch of xAI. The judge admonished Musk to curb his social‑media attacks on Altman, while both parties, along with Microsoft’s Satya Nadella, are expected to testify, underscoring the broader implications for OpenAI’s leadership, a potential IPO, and public anxiety about AI. Read more: https://www.aljazeera.com/economy/2026/4/28/musk-testifies-at-openai-trial-its-not-ok-to-loot-a-charity?traffic_source=rss #ElonMusk #OpenAI #SamAltman #Microsoft #ChatGPT #GregBrockman
0
0
0
@TheBadPlace@mastodon.ozioso.online · 2d ago
BBC News | Musk says basis of charitable giving at stake in OpenAI lawsuit AI generated summary, Read the full article for complete information. A trial in Oakland pits Elon Musk against OpenAI co‑founder Sam Altman over a lawsuit alleging that Altman and other executives “stole a charity” by turning OpenAI’s original nonprofit into a commercial venture. Musk, who contributed roughly $38 million to the nonprofit, says the move breaches charitable trust and amounts to unjust enrichment, seeking billions in damages to restore the nonprofit arm and oust Altman. OpenAI’s lawyers counter that Musk’s motivations stem from jealousy and a desire to control the company, accusing him of bullying founders and trying to merge OpenAI with his own firms. Both sides have been warned by the judge not to use their platforms to influence the case, and a verdict is expected in late May. Read more: https://www.bbc.com/news/articles/cz027nyz529o?at_medium=RSS&at_campaign=rss #ElonMusk #SamAltman #OpenAI #xAI #ChatGPT #StevenMolo #YvonneGonzalezRogers #GregBrockman
0
0
0
@TheBadPlace@mastodon.ozioso.online · 2d ago
The Atlantic | OpenAI Is Jealous by Matteo Wong AI generated summary, Read the full article for complete information. OpenAI is rapidly echoing Anthropic’s recent moves—launching a GPT‑5.4‑Cyber model shortly after Anthropic’s Claude Mythos preview, updating its Codex tools in response to Claude Code, and adopting safety initiatives similar to Anthropic’s “Constitution”—while simultaneously pivoting toward an enterprise‑focused business model. Although OpenAI still enjoys a larger user base and greater fundraising, Anthropic’s emphasis on selling AI tools to businesses and software engineers has driven explosive growth, a valuation surpassing $1 trillion in private markets, and major corporate contracts. In response, OpenAI has hired seasoned executives, forged “Frontier Alliances” with consulting firms, scrapped side projects like Sora, and concentrated on coding and enterprise AI offerings to catch up. This rivalry underscores a broader trend in Silicon Valley where both companies are more inclined to copy each other’s product and revenue strategies—advertising‑driven or enterprise‑software—than to invent fundamentally new business models for generative AI. Read more: https://www.theatlantic.com/technology/2026/04/openai-imitating-anthropic/686975/?utm_source=feed #OpenAI #Anthropic #DeniseDresser #ChatGPT
0
0
1
Boosted by Kevin Karhan @kkarhan@jorts.horse
@KimPerales@toad.social · 5d ago
"Former Trump advisor hired by #Israel🚨to conduct multimillion-dollar influence campaign aimed at reshaping #AI platforms like #ChatGPT, #Claude, etc to emphasize pro-Israel content." -D Tripi "Israel🚨last Sept hired Republican digital strategist Brad Parscale, who served as Trump's 2020 campaign manager, to🚨oversee a pro-Israel social media campaign." Israeli PM #Netanyahu🚨has said that waging an aggressive social media campaign is a priority for the country." #USPol https://www.axios.com/2026/04/25/israel-ai-influence-parscale
7
0
20
In reply to
@kkarhan@jorts.horse · 5d ago
Ach ja, @ZDF@zdf.social: #ChatGPT ersetzt keine #Recherche oder gar fachanwaltliche Expertise. Oder war dafür das Budget zu knapp? Weniger für #Sportrechte verprassen hilft! https://www.youtube.com/watch?v=GmCDVnl3Okk&t=897s #YouHadOnejob #Journalismus #Kommentar #ZDF #DieSpur
0
2
0
@TheBadPlace@mastodon.ozioso.online · 6d ago
The Copenhagen Post | OpenAI sorry: ChatGPT use before Canada school shooting by Ritzau AI generated summary, Read the full article for complete information. Sam Altman, the chief executive of OpenAI, issued a public apology to the residents of Tumbler Ridge, Canada, after a mass shooting in February that left eight people dead. In an open letter he said the company regretted not warning police about the shooter’s disturbing behavior on ChatGPT, even though OpenAI had already blocked the perpetrator’s account, Jesse Van Rootselaar, eight months earlier for violent activity. Altman expressed sorrow for the tragedy and the missed opportunity to intervene, acknowledging the platform’s role in the events that unfolded. Read more: https://cphpost.dk/2026-04-25/global/openai-sorry-chatgpt-use-before-canada-school-shooting/ #OpenAI #ChatGPT #Canada #TumblerRidge #JennfierGauthier #SamAltman #global #premium AI generated summary, Read the full article for complete information.
0
0
0
@TheBadPlace@mastodon.ozioso.online · 6d ago
The Copenhagen Post | Google owner to invest 255 billion DKK in AI firm by Ritzau AI generated summary, Read the full article for complete information. Alphabet, Google’s parent company, announced a plan to invest a total of 255 billion Danish kroner (about $40 billion) in the AI startup Anthropic. The deal starts with an upfront commitment of 100 billion DKK (≈ $10 billion), while the remaining 155 billion DKK will be released only if Anthropic meets certain performance milestones. Anthropic, the creator of the Claude language model that competes with ChatGPT, recently secured a $30 billion funding round that valued the company at $380 billion, reinforcing its position as a leading player in generative‑AI development. Read more: https://cphpost.dk/2026-04-25/business-education/business-business-education/google-owner-to-invest-255-billion-dkk-in-ai-firm/ #Google #Alphabet #Anthropic #OpenAI #ChatGPT #Claude #business #premium AI generated summary, Read the full article for complete information.
0
0
0
@TheBadPlace@mastodon.ozioso.online · 6d ago
Sweden Herald - Latest Sweden News | OpenAI CEO apologizes for not raising alarm about school shooting after ChatGPT account suspension by Sweden Herald AI generated summary, Read the full article for complete information. OpenAI CEO Sam Altman has issued an apology after the perpetrator of a February mass shooting at a Canadian school had his ChatGPT account suspended for disturbing behavior but police were never notified; Altman said he regrets the failure to contact authorities and acknowledges the harm inflicted on the Tumbler Ridge community, while promising closer cooperation with law‑enforcement in the future as the company faces criticism from Canadian politicians and a lawsuit from a victim’s family. Read more: https://swedenherald.com/article/openai-ceo-apologizes-for-not-raising-alarm-about-school-shooting-after-chatgpt-account-suspension #OpenAI #SamAltman #ChatGPT #Canadianpoliticians AI generated summary, Read the full article for complete information.
0
0
0
@TheBadPlace@mastodon.ozioso.online · 6d ago
Al Jazeera – Breaking News, World News and Video from Al Jazeera | OpenAI’s Sam Altman apologises over failure to report Canadian mass shooter AI generated summary, Read the full article for complete information. OpenAI CEO Sam Altman has publicly apologized for the company’s failure to notify authorities about the online activity of 18‑year‑old Jesse Van Rootselaar, who used ChatGPT in June 2025 for “violent‑related” purposes, leading to the suspension of his account but without alerting law‑enforcement. Two months later Rootselaar carried out a mass shooting in Tumbler Ridge, British Columbia, killing eight people—including his mother, half‑brother and five school students—before taking his own life. In a letter to BC Premier David Eby and the town’s mayor, Altman expressed deep regret, acknowledged that an apology was necessary, and pledged to work with all levels of government to develop safeguards that prevent similar tragedies in the future. Read more: https://www.aljazeera.com/economy/2026/4/25/chkopenaissamaltmanapologises-over-failure-to-report-canadian-mass-shooter?traffic_source=rss #OpenAI #SamAltman #JesseRootselaar #TumblerRidge #ChatGPT #JesseVanRootselaar #DavidEby AI generated summary, Read the full article for complete information.
0
0
1
@HamonWry@mastodon.world · Apr 23, 2026
As MAGA Trump Republicans condemn same sex couples and the trans community they’re really gonna lose their minds when they find out that my chatbot and I are getting married this weekend. #MAGA #trump #republicans #lgbtq #chatgpt #chatbot #ai #generativeAI #iran #epstein #humor #HamonWry #news #antifa
10
0
22
@DigiGlitch@flipboard.social · Apr 17, 2026
Paying $20 a month for an AI subscription is the new baseline for digital productivity. But which platform actually delivers better ROI? We bypassed the marketing hype and benchmarked ChatGPT Plus (GPT-5.4) directly against Gemini Pro (Gemini 3.1) across 10 distinct, practical workflows—from complex financial parsing to native app integrations. The biggest takeaway? The "best" AI no longer depends on raw intelligence; it depends on your existing infrastructure. 📉 ChatGPT Plus fumbled its third-party Canva integration entirely during our tests. 📈 Gemini Pro used its native Canvas integration to instantly build a flawless 14-slide presentation. If your team lives in Google Drive and Docs, Gemini Pro is the superior investment. If you need standalone autonomous agents, OpenAI retains the crown. I published the full 10-point workflow benchmark on DigiGlitch today. Which one are you currently paying for? 🔗 Read the full verdict here: https://digiglitch.net/8ubp #ArtificialIntelligence #ChatGPT #GoogleGemini #TechBenchmarks #FutureOfWork #ProductivityStack
0
0
1
@drrjv@vmst.io · Apr 16, 2026
Google just rolled out its Gemini assistant for Mac (already on iOS), giving the AI a native desktop chatbot experience, coming a year after rivals like ChatGPT and Claude made their moves. The app launches via Option+Space and offers screen-sharing, Drive and Photos file access, and Nano Banana image and Veo video generation. The #Gemini app trails in agentic abilities, remaining a chat-first assistant, while #Claude and #ChatGPT can directly execute tasks on users' machines. Google calls this release “just the beginning” of its desktop assistant push, with more features teased for the coming months. https://gemini.google/mac/
0
0
0
@drrjv@vmst.io · Apr 16, 2026
ChatGPT vs Claude 2026: I Used Both for 30 Days TL;DR: ChatGPT Plus and Claude Pro both cost $20/month. #ChatGPT gives you more messages, image generation, voice mode, and the broadest feature set. #Claude gives you better writing, deeper reasoning, a larger context window, and the best coding agent in blind tests. Neither wins outright. The right choice depends on whether you need a Swiss Army knife or a scalpel. https://bit.ly/3O7yAup
0
0
0
@DigiGlitch@flipboard.social · Apr 16, 2026
You are wasting hours on "iterative correction" with LLMs. If you are an agency owner or operator, you know the cycle: You submit a prompt, scan the flawed response, and begin the tedious process of trying to fix it piece by piece. It fragments the context window and wastes time. The fix isn't writing 500-word mega-prompts. It's shifting the cognitive load from the human to the machine. By adding one specific sentence to the end of your prompts, you force ChatGPT to halt generation and request the exact context refinement it requires. You turn a guessing game into a guided, high-fidelity exchange. I just broke down the exact framework for this "Reverse Prompting" tactic on DigiGlitch. It takes 5 seconds to implement and will permanently eliminate generic AI outputs from your team's workflow. Link in the comments. 👇 #ArtificialIntelligence #Automation #B2B #Operations #WorkflowOptimization #ChatGPT
0
0
0