OpenAI accused of showing misleading charts during GPT-5 launch, Sam Altman calls it ‘mega screwup'
In one of the charts presented during the live-stream, it shows GPT-5 supposed prowress compared to OpenAI's o3 in different parameters. In of the parameters called 'coding deception', GPT-5 ges a 50 percent deception rate compared to o3's 47.4 score, a pretty close call statistically speaking but the graph shows almost the opposite. In stead of showing a higher bar for GPT-5, OpenAI showed off o3 with a much bigger bar.
The company went on to correct the mistake in a subsequent blogpost but the deception rate for GPT-5 there is certified as 16.5 percent.
As if that wasn't enough, another chart from the live-stream comparing GPT-5 to o3 and GPT-4o shows the new model with a score of 74.9 compared to 69.1 and 30.8 for the other two models respectively. While o3 and GPT-4o have a vast difference in scores, the chart shows them with almost the same length bar compared to a bigger bar for GPT-5.
OpenAI CEO Sam Altman commented on the wrong charts shows during the live-stream, calling it a 'mega chart screwup' while stating that a correct version of the charts have been uploaded on OpenAI's blog post.
Meanwhile, an OpenAI marketing staff also apologized for the error in a post on X (formerly Twitter), writing, 'We fixed the chart in the blog guys, apologies for the unintentional chart crime 🙏'
OpenAI says GPT-5 comes with major improvements in areas like accuracy, speed, reasoning, context recognition, structured thinking, and problem-solving compared to the company's GPT-4o model.
The biggest change with GPT-5 is the introduction of a unified system with an 'efficient model' powering GPT-5's normal tasks while a dedicated reasoning model called GPT-5 Thinking handling harder reasoning based tasks.
Unlike in the past, where users had to choose the model for each query, GPT-5 features a real-time router trained on real signals to instantly decide which model to use.
The new model comes with major improvements in coding related tasks and also possesses the ability to create apps, games and websites using natural language prompts. There is also claimed to be a marked improvement in writing and health related tasks.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Mint
7 hours ago
- Mint
Billions flow to new hedge funds focused on AI-related bets
Leopold Aschenbrenner emerged last year as a precocious artificial-intelligence influencer after publishing a widely read manifesto. Then he decided to try his hand at stock picking. The 23-year-old with no professional investing experience quickly raised more money for a hedge fund than most pedigreed portfolio managers can when they strike out on their own. As valuations of Nvidia, OpenAI and other artificial-intelligence companies continue to soar, so do investments in hedge funds hoping to ride the AI wave. Aschenbrenner's San Francisco-based firm, Situational Awareness, now manages more than $1.5 billion, according to people familiar with the matter. He has described the firm as a 'brain trust on AI." His strategy involves betting on global stocks that stand to benefit from the development of AI technology, such as semiconductor, infrastructure and power companies, along with investments in a few startups, including Anthropic. He told investors he plans to offset those with smaller short bets on industries that could get left behind. Situational Awareness gained 47% after fees in the first half of the year, one of the people said. In the same period, the S&P 500 gained about 6%, including dividends, while an index of tech hedge funds compiled by research firm PivotalPath gained about 7%. Aschenbrenner, a native of Germany, briefly worked as a researcher at OpenAI before being pushed out. He named Situational Awareness after the 165-page essay he wrote about the promise and risks of artificial superintelligence. He recruited Carl Shulman, another AI intellectual who used to work at Peter Thiel's macro hedge fund, as director of research. The firm's backers include Patrick and John Collison, the billionaire brothers who founded payments company Stripe, as well as Daniel Gross and Nat Friedman, whom Mark Zuckerberg recently recruited to help run Meta's AI efforts. Graham Duncan, a well-known investor who organizes the Sohn Investment Conference, is an adviser. 'We're going to have way more situational awareness than any of the people who manage money in New York," Aschenbrenner told podcaster Dwarkesh Patel last year. 'We're definitely going to do great on investing." In another sign of the demand for Aschenbrenner's services, many investors agreed to lock up their money with him for years. Other recent launches include an AI-focused hedge fund from Value Aligned Research Advisors, a Princeton, N.J.-based investment firm founded by former quants Ben Hoskin and David Field. The fund, launched in March, has already amassed about $1 billion in assets, a person familiar with it said. VAR also manages about $2 billion in other AI-focused investment strategies. VAR's investors have included the philanthropic foundation of Facebook co-founder Dustin Moskovitz, according to regulatory filings reviewed by fund-data tracker Old Well Labs. Veteran hedge-fund firms are entering the fray, too. Last year, Steve Cohen tapped one of his portfolio managers at Point72 Asset Management, Eric Sanchez, to start an AI-focused hedge fund that Cohen planned to stake with $150 million of his own money. Assets at the fund, called Turion—after AI theorist Alan Turing—now exceed $2 billion, people familiar with the matter said. Turion is up about 11% this year through July after it gained about 7% last month, the people said. It is no surprise that thematic funds are springing up to capitalize on the AI frenzy. In years past, hedge funds that specialized in the transition to clean energy and investing with an environmental, social and corporate-governance lens proliferated in response to client demand. Identifying a winning theme isn't the same thing as trading it well. Investors' tastes can be fickle; many prominent ESG hedge funds have either shrunk or gone out of business. The market swoon that followed the January release of an advanced, low-cost language model from Chinese company DeepSeek showed the fragility of the valuations of AI winners, though the market has roared back since then. AI-focused investors argue the long-term trend of development and adoption are inevitable, even if there are bumps along the way. With only so many publicly traded companies that operate in the AI-adjacent economy today, stock picking funds often pile into the same positions as one another and more generalist hedge funds. Vistra, a power producer that supplies the juice to AI data centers, was a top-three U.S. position of both Situational Awareness and VAR Advisors as of March 31, according to their most recent securities filings. Other hedge-fund managers are debuting funds to make investments in privately held AI companies and startups. Gavin Baker's Atreides Management teamed up with Valor Equity Partners to launch a venture-capital fund earlier this year that has raised millions from investors including Oman's sovereign-wealth fund. Each firm separately invested in Elon Musk's xAI. At least one portfolio manager is planning an AI hedge fund as a comeback vehicle. Sean Ma wound down his Hong Kong-based firm, Snow Lake Capital, after it agreed to pay about $2.8 million to settle Securities and Exchange Commission charges last year that the firm participated in stock offerings of companies that it had also bet against. Ma took over an investment firm called M37 Management in Menlo Park, Calif., earlier this year. He is currently fundraising for a hedge fund focused on AI software and hardware.
Economic Times
8 hours ago
- Economic Times
Can Trump's Fannie Mae and Freddie Mac IPO plan slash mortgage rates? Bill Ackman says...
Synopsis Donald Trump is reportedly planning to IPO Fannie Mae and Freddie Mac, potentially the largest IPO in history. Billionaire Investor Bill Ackman suggests merging the two mortgage giants to reduce mortgage rates and government oversight costs. Ackman believes privatization could yield substantial gains for the government, citing their improved capitalization and government backing. AP American hedge fund manager Bill Ackman took to X, formerly known as Twitter, and one way to reduce mortgage rates would be to merge government-sponsored enterprises Fannie Mae and Freddie Mac US President Donald Trump Saturday seemed to acknowledge reporting by The Wall Street Journal on Friday that he plans to IPO Fannie Mae and Freddie Mac by the end of this year. The President and his economic advisers are planning a historic sale of stock in Fannie Mae and Freddie Mac, the government-owned mortgage giants that help provide stability and affordability to America's home loan market. Reacting to the development, American hedge fund manager Bill Ackman took to X, formerly known as Twitter, and one way to reduce mortgage rates would be to merge government-sponsored enterprises Fannie Mae and Freddie Mac. He suggested the merger move would help reduce mortgage rates and achieve huge synergies both in their operations and in the trading price. ALSO READ: 'Ban Gay sex, end women's voting': Pete Hegseth sparks controversy for re-posting pastor's radical message Bill Ackman said Fannie and Freddie merger would also reduce the costs and risks of government way to reduce mortgage rates would be to merge Fannie and Freddie. A merger would enable them to achieve huge synergies both in their operations and in the trading price and spreads of their MBS, savings which could be passed along to consumers in the form of reduced mortgage rates, Ackman wrote in his post. "A merger would also reduce the cost and risks of government oversight as there would be only one institution that would require FHFA oversight. I suspect that this is @realDonaldTrump 's idea as implied by his post below. It's a really good one," his post read. US-government owned twin giants, Fannie Mae and Freddie Mac are tasked with expanding credit availability in the American market by securitising mortgages. Their shares surged over 20 per cent on Friday after the Wall Street Journal reported that the Trump administration may privatise the two institutions this year. ALSO READ: Powerball jackpot rises to $479 million: Who won lottery jackpot last night? Lotto results, drawing time US President Donald Trump has previously met the top leadership of US investment banks such as Citigroup, the Bank of America, Goldman Sachs and JPMorgan Chase to explore potential public offerings of the twin mortgage giants, Reuters reported, citing an the plans have not been finalised yet, and Trump continues to weigh various options, according to a senior administration official. But the White House believes an initial public offering of up to 15% of the two companies' shares could raise $30 billion, which could make it the largest IPO in has been weighing an IPO for years now. During his first term, Trump attempted — but ultimately failed — to privatize Fannie Mae and Freddie Mac, removing them from government conservatorship. Now, in his second term, he has revived the push. In May, he wrote on Truth Social that he was 'giving very serious consideration to bringing Fannie Mae and Freddie Mac public,' adding that he would consult with his Cabinet before making a decision 'in the near future.'Trump has argued for the monetisation of these two institutions, which were brought under US government control in the aftermath of the 2008 financial crisis. In May this year, Trump floated the idea while emphasising that the government will maintain its implicit guarantees for the securities issued by the two institutions. ALSO READ: Last planet parade of 2025 happening today? How to watch the rare planetary alignment in the US Trump backer Bill Ackman, a long-time shareholder in the twin behemoths, has repeatedly called for their privatisation. Ackman, founder, Pershing Capital Management, told Forbes magazine last month that the US government is the preferred stockholder of the twins, and in a position to realise gains worth $300 billion. He argued that the two institutions were 'vastly better capitalised' today than for the past 60 two institutions are not banks, but tap creditworthy mortgage buyers and pack the mortgages in securities to be sold on the market, Ackman explained. Fannie Mae and Freddie Mac have guarantees worth $7 trillion coupled with enormous cash flows, apart from a government backing, underlining their ability to weather any future crisis, Ackman added.
Time of India
9 hours ago
- Time of India
OpenAI bans ChatGPT from answering breakup questions; Sam Altman calls the new update 'annoying'
OpenAI is adjusting ChatGPT's approach to sensitive emotional queries, shifting from direct advice to facilitating user self-reflection. This change addresses concerns about AI's impact on mental well-being and aims to provide more thoughtful support. OpenAI is consulting experts and implementing safeguards like screen-time reminders and distress detection to ensure responsible AI interaction. Nowadays, many of us have turned AI platforms into a quick guidance source for everything from code to personal advice. But as artificial intelligence becomes a greater part of our emotional lives, companies are becoming aware of the risks of over-reliance on it. But can a chatbot truly understand matters of the heart and emotions? With growing concerns about how AI might affect mental well‑being, OpenAI is making a thoughtful shift in how ChatGPT handles sensitive personal topics. Rather than giving direct solutions to tough emotional questions, the AI will now help users reflect on their feelings and come to their own conclusions. OpenAI has come up with significant changes OpenAI has announced a significant change to how ChatGPT handles relationship questions. Instead of offering direct answers like 'Yes, break up,' the AI will now help users think through their dilemmas by giving self-reflection and weighing pros and cons, particularly for high-stakes personal issues. This comes as there have been several issues over AI getting too direct in emotionally sensitive areas. According to reports from The Guardian, OpenAI stated, 'When you ask something like: 'Should I break up with my boyfriend?' ChatGPT shouldn't give you an answer. It should help you think it through—asking questions, weighing pros and cons.' The company has also said that 'new behaviour for high‑stakes personal decisions is rolling out soon. We'll keep tuning when and how they show up so they feel natural and helpful,' according to OpenAI's statement via The Guardian. To ensure this isn't just window dressing, OpenAI is gathering an advisory group of experts in human-computer interaction, youth development, and mental health. The company said in a blog post, 'We hold ourselves to one test: if someone we love turned to ChatGPT for support, would we feel reassured? Getting to an unequivocal 'yes' is our work' OpenAI CEO says that the new update is.... This change follows user complaints about ChatGPT's earlier personality tweaks. According to the Guardian, CEO Sam Altman admitted that recent updates made the bot 'too sycophant‑y and annoying.' He said, 'The last couple of GPT‑4o updates have made the personality too sycophant‑y and annoying (even though there are some very good parts of it), and we are working on fixes asap, some today and some this week.' Altman also teased future options for users to choose different personality modes OpenAI is also implementing mental health safeguards. Updates will include screen-time reminders during long sessions, better detection of emotional distress, and links to trusted support when needed
