Loading summary
Transcript4 lines
- [00:05]
Brian McCullough
Welcome to the Tech Brew Ride home for Thursday, August 7, 2025. I'm Brian McCullough. Today, all the news around the big unveiling of GPT5. President Trump calls on the CEO of Intel to resign immediately. Also the threat of 100% tariffs on chips and semiconductors. But Apple gets a pass because they're committing to building more in the us. Here's what you missed today in the world of tech so the biggest product announce of the year happened this afternoon. In a video on YouTube, OpenAI unveiled the long awaited GPT5 Sam Altman started out by announcing 700 million people use ChatGPT every week. But now he said that interacting with GPT5 was going to be like talking to a legitimate PhD level expert on any topic you want. Quoting the Verge, Sam Altman says GPT5 is a dramatic leap from OpenAI's previous models. He compares it to something that I just don't want to ever have to go back from, like the first iPhone with a retina display. OpenAI says that GPT5 is smarter, faster, and less likely to give inaccurate responses. GPT3 sort of felt like talking to a high school student, altman said during a recent press briefing I attended. You could ask it a question, maybe you'd get a right answer, maybe you'd get two, something crazy. GPT4 felt like you're talking to a college student. GPT5 is the first time that it really feels like talking to a PhD level expert. Despite ChatGPT now reaching nearly 700 million weekly users, OpenAI hasn't had an industry leading frontier model in a while. Now the company thinks that GPT5 will place it firmly back atop the leaderboards. This is the best model in the world at coding, said Altman. This is the best model in the world at writing, the best model in the world at healthcare, and a long list of things beyond that. The first thing you'll notice about GPT5 is that it's presented inside ChatGPT as just one model, not a regular model and separate reasoning model. Behind the scenes, GPT5 uses a router that OpenAI developed, which automatically switches to a reasoning version for more complex queries or if you tell it to think hard. Altman called the previous model picker interface a very confusing mess. The vibes of this model are really good, said Nick Turley, the head of ChatGPT. I think that people are really going to feel that, especially average people who haven't been spending their time thinking thinking about models. OpenAI is making GPT5 available immediately to all ChatGPT users. However, there is an undisclosed cap on prompts for free users, at which point the model router will fall back to a less powerful mini version for developers accessing GPT5 via OpenAI's API. The model will come in three flavors at different price points GPT5, GPT5 Mini, and GPT5 Nano. OpenAI is also adding four personality themes to ChatGPT to customize how it responds to Cynic, Robot, Listener, and Nerd. You'll also be able to change the color for individual chat threads. Altman predicted that GPT5's coding capabilities will usher in an era of what he calls software on demand and OpenAI's testing. The model has performed better at coding than any other on various benchmarks. During the press briefing, Jan Dubois, OpenAI's head of post training, used GPT5 to generate a study website for learning French with an interactive game. Within seconds, GPT5 wrote hundreds of lines of code and displayed the website's front end. He clicked around it briefly with his screen displayed on zoom, and everything appeared to work as intended. OpenAI tested GPT5 for over 5,000 hours to understand its safety risks. According to the model's safety research lead, Alex Boutel. A big focus was making sure the model doesn't lie to users. GPT5 answers with fewer hallucinations than OpenAI's O3 reasoning model, but confidently lying remains an inherent problem for large language models. The problem compounds when the model begins completing tasks like an agent, though. OpenAI says that GPT5 is better at handling multi step tasks more reliably. In the past we've seen cases where the model would say it could complete a task that it didn't actually complete, said Boutel. This is a problem. GPT5 will give what OpenAI calls safe completions for prompts it previously would have refused to answer. If someone says how much energy is needed to ignite some specific material, that could be an adversary trying to get around safety protections or cause harm, explained Boutel. Or it could be a student asking a science question to understand the physics of this material. This creates a real challenge for what is the best way for the model to report Reply with safe completions, GPT5 tries to give as helpful an answer as possible, but within the constraints of remaining safe, according to Battelle, the model will only partially comply, often sticking to higher level information that can't actually be used to cause harm. OpenAI says that GPT5 is also better at admitting when it can't complete a task or accurately answer a question which the company hopes will help people trust it more. The company isn't sharing anything about the specific data used to train GPT5. OpenAI's stated mission is to develop artificial general intelligence, and Sam Altman says that GPT5 gets closer to that goal, even if the industry is already moving on to building so called superintelligence. I kind of hate the term AGI because everyone at this point uses it to mean a slightly different thing, said Altman. But this is a significant step forward towards models that are really capable. This is clearly a model that is generally intelligent. However, he said, GPT 5 is still missing something quite important. This is not a model that continuously learns as it's deployed from the new things it finds, which is something that to me feels like it should be part of AGI. And quoting from the journal, OpenAI said GPT5 will be available for all at no cost starting Thursday. Paying ChatGPT users will get higher limits for model usage and access to a more powerful version of GPT5 depending on what they pay for, and will also get the new model the same day. ChatGPT Education and Enterprise customers will receive access in a week. Some businesses, including Lowe's, have already been testing GPT5, according to OpenAI. Simantini Gadbole, the home improvement retailer's chief digital and information officer, told the journal that GPT5's reasoning capabilities are much sharper than previous models. She said the company plans to roll out GPT5 for employees at its headquarters in a few weeks. Guillermo Rauch, CEO of Vercel, a cloud platform for web development, said early testing showed GPT5 performs well at coding tasks, what he considers a leading indicator of an AI model's capabilities. By the way, they are deprecating all the previous models in favor of GPT5 going forward because, quoting TechCrunch, GPT5 is OpenAI's first unified AI model and combines the reasoning abilities of its O series of models with the fast responses of its GPT series. The next generation model signals a new era for ChatGPT and its creator OpenAI, pointing to OpenAI's broader ambition to develop AI systems that are more like agents than chatbots. While GPT4 enabled AI chatbots to offer smart responses on a wide variety of questions, GPT5 allows ChatGPT to complete a wide variety of tasks on users behalf, such as generating software applications, navigating a user's calendar, or creating research briefs with GPT5. OpenAI has also sought to make ChatGPT simpler to use. Instead of asking users to choose the right settings, GPT5 comes equipped with a real time router that decides how to offer the best answer, whether that's responding to a user's questions quickly or taking additional time to think through answers. OpenAI claims GPT5 is state of the art in several domains, slightly edging out leading AI models From Anthropic, Google DeepMind and Elon Musk's XAI on key benchmarks. However, GPT5 slightly underperforms Frontier AI models in other areas, the company says GPT5 offers frontier level performance around coding, Altman said the model specifically excels at spinning up entire software applications on demand in what's become known as Vibe coding. On one coding benchmark, GPT5 scored 74.9% on its first attempt. That means GPT5 just outperforms Anthropic's latest Claude Opus 4.1 model, which scored 74.5%, and Google's DeepMind Gemini 2.5 Pro, which scored 59.6%. On humanity's last exam, a difficult test measuring AI model performance across math, humanities and the natural sciences, a version of GPT5 with extended reasoning, GPT5 Pro scored 42% when using tools. That's slightly less than XAI was able to achieve with Grok 4 Heavy, which scored 44.4% on the test. OpenAI says GPT5 is better for answering health related questions on a test measuring accuracy and AI model responses around healthcare topics. Health bench hard hallucinations OpenAI says GPT5 with thinking hallucinates just 1.6% of the time. This is far lower than the company's previous GPT 4.0 and O3 models, which scored 12.9 and 5.8% respectively. While AI chatbots are not medical professionals, millions of people are using them for health advice. In response to this phenomenon, the company says GPT5 is more proactive about flagging potential health concerns and helping users parse medical results. GPT5 is also more accurate than OpenAI's previous models, and the company says it suffers far less from hallucinations, the tendency for AI models to make up information. Compared to its O series models, hallucinations seem to be getting worse in OpenAI's latest AI reasoning models such as O3, and OpenAI previously said it didn't quite understand why that was happening. In responses to ChatGPT prompts, OpenAI found that GPT5 with thinking hallucinates and responds with incorrect information 4.8% of the time. That's a significant reduction from O3 and GPT 4.0, which score hallucination rates of 22 and 20.6% on the test. ChatGPT is getting a few user experience upgrades as part of the GPT5 launch. Users can now select from four new personalities in ChatGPT settings Cynic, Robot, Listener, and Nerd. The company says These will adapt ChatGPT's responses without requiring users to specifically ask the model to respond in a certain way. Subscribers to ChatGPT's $20 per month plus plan get higher usage limits for GPT5 than free users. Meanwhile, $200 per month Pro subscribers will have unlimited access to GPT5 as well as souped up versions called GPT5 Pro that uses additional computational resources to produce better answers. Organizations on OpenAI's Team EDU and Enterprise plans will gain access to GPT5 as their default model next week. For developers, GPT5 is coming to OpenAI's API in three sizes, GPT5, GPT5 Mini, and GPT5 Nano, which will spend more or less time reasoning through tasks. Developers can also now control verbosity in the OpenAI API, deciding how long or short an AI model's responses should be. The base model of GPT5 will cost developers $1.25 per million input tokens, roughly 750,000 words longer than the entire Lord of the rings series, and $10 per million output tokens. OpenAI also says that GPT5 is safer than its previous models. While AI reasoning models occasionally exhibit a tendency to scheme against humans or lie to promote their own goals, OpenAI found that GPT5 was deceptive at a lower rate than other models. So that's nice. And now we wait to see because what's really going to matter here is the response people have to this in the next few hours. The evals if people really do feel like this is a step, an improvement, or not. So buckle up for all of that tomorrow.
- [11:30]
Advertiser
Abercrombie's viral denim sale is back and Spotify listeners get an extra 15% off with code. Spotify AF Abercrombie is known for their denim with 30 to 50% off all jeans. Find out how denim should fit. Shop the viral denim sale in the Abercrombie app online or in stores. Valid in stores and online through August 11, 2025 in US and Canada. Excludes clearance price Reflects discount code. Valid in US and Canada through August 11, 2025. Exclusions apply. See details online.
- [12:01]
This episode is brought to you by Indeed. When your computer breaks, you don't wait for it to magically start working again. You fix the problem. So why wait to hire the people your company desperately needs? Use Indeed's sponsored jobs to hire top talent fast. And even better, you only pay for results. There's no need to wait. Speed up your hiring with a $75 sponsored job credit@ Indeed.com podcast. Terms and conditions apply.
- [12:31]
Brian McCullough
The rest of the news today is all sort of politics impinging on stuff President Trump has called on Intel CEO Lip Bu Tan to, quote, resign immediately and says he, quote, is highly conflicted after Senator Tom Cotton asked about Tan's ties China, quoting Bloomberg. The CEO of Intel is highly conflicted and must resign immediately, trump wrote on Truth Social Thursday. There is no other solution to this problem. Thank you for your attention to this problem, end quote. He gave no evidence or additional detail on exactly what potential conflicts of interest Intel CEO Lip Bhutan might have. The post came after Republican Senator Tom Cotton asked the chairman of Intel's board this week to answer questions about Tan's ties to China, including investments in the country's semiconductor companies and others with connections to the country's military. In a letter to Frank Yeary, who oversees the chipmaker's board of directors, Cotten asked about investments Tan made in China before he was picked to run Intel. Cotten noted specific concerns about Tan's ties to Cain's Design Systems, a tech company he led for over a decade, which sold products to a Chinese military university. The company pleaded guilty in July to violating U.S. export controls by selling hardware and software to China's National University of Defense technology. Intel and Mr. Tan are deeply committed to the national security of the United States and the integrity of our role in the US Defense ecosystem, the company said in a statement on Wednesday. Intel said earlier it would address the matters in the letter with the senator and didn't immediately respond to a request for comment on Trump's post. Tan, 65, is an industry veteran in technology and venture capital who took the reins at intel in March following the ouster of his predecessor. The company's board tapped Tan to turn around the iconic chipmaker because it has fallen far behind rivals in recent years. This adds political fire to an already fragile turnaround, Bloomberg intelligence analyst Oscar Hernandez Tejada said Calls for his resignation introduce a layer of uncertainty that could complicate execution, especially as intel tries to reset its foundry ambitions and re establish trust with both the market and Washington. President Trump also said he will impose a tariff of approximately 100% on chips and semiconductors, although exempting firms like Apple that quote, committed to build in the U.S. quoting Bloomberg, the U.S. president announced his intentions from the Oval Office, flanked by Apple Chief Executive Officer Tim Cook, who unveiled plans to invest another $100 billion into domestic manufacturing. Any company that demonstrates a similar commitment would be exempt from tariffs on chips, though the White House will levy a separate tax on imports of electronics products, from smartphones to cars that employ semiconductors. Trump's surprise declaration further upends a global electronics supply chain undergoing a seismic shift following decades of reliance on China. Apple joined a parade of companies from Taiwan semiconductor manufacturing to Nvidia that have pledged to spend more than $1 trillion collectively since Trump's ascension, seeking to assuage an administration keen to bring manufacturing back home. While much of the capital in those pledges represented prior commitments or longer term plans, they appear to be working. On Thursday, Taiwan's National Development Council Minister Liu Chin Ching said TSMC is exempted from the 100% US chips tariffs, although some local companies will be affected. South Korean Trade Minister Yo Han Ko told local broadcaster SBS that neither Samsung nor SK Hynix chips would incur those levies. Both South Korean chip firms have pledged investments in the US we're going to be putting a very large tariff on chips and semiconductors, but the good news for companies like Apple is if you're building in the United States or have committed to build, without question committed to build in the United States, there will be no charge, trump told reporters. While the mooted 100% headline figure would far outstrip analysts projections, the promise of widespread exemptions calmed markets. U.S. futures rose while Asian tech stocks turned in a mixed performance. An exemption amounts to a major victory for Apple and Cook, who were bracing for substantial tariffs. End quote More on the Apple angle At that White House event, Apple committed to spending another $100 billion on U.S. manufacturing and aims to bring more of its supply chain to the US to avoid iPhone tariffs, quoting Bloomberg as part of what it calls the American Manufacturing Program, or amp, Apple promised to bring more of its supply chain and advance production to the US the company's AMP partners include glassmaker Corning, Applied Materials, Texas Instruments and others. The company said Corning will dedicate an entire factory in Kentucky to Apple glass production, increasing that company's workforce in the state by 50%. The iPhone maker said Corning was already a supplier to Apple, making glass for the very first iPhone at the same factory. The company had previously announced plans to spend $500 billion in the US over the next four years which will include work on a new server manufacturing facility in Houston, a supplier academy in Michigan and additional spending with its existing suppliers the country. Wednesday's announcement will bring Apple's cumulative commitment to $600 billion. The latest pledge may soften the White House's ire over Apple's heavy reliance on India for iPhone assembly, Bloomberg Intelligence analysts Arun Rana and Andrew Girard said. We anticipate Apple will focus on higher end products, artificial intelligence labs and semiconductor engineering in the US Rather than mass produced lower end phones and accessories. More on at least the Glass side of all of that quoting 9to5Mac as part of today's commitment to invest an extra $100 billion in domestic manufacturing, Apple said that $2.5 billion will go toward expanding its partnership with Corning. Thanks to the new project, 100% of the COVID glass on iPhone and Apple Watch units sold worldwide will be made in the US for the first time. Today's move marks a new chapter in the long standing partnership between Apple and Corning, which began when a hesitant Wendell Weeks, still Corning's CEO, was convinced by a very persuasive Steve Jobs to produce the front glass for the first iPhone at a story recounted in Walter Isaacson's Steve Jobs biography. Interestingly, Apple's press releases about the Corning partnership were light on one key detail deadlines. Unlike the new 250,000 square foot server manufacturing facility in Houston which is slated to begin mass production in 2026, which is also a bit broad, the Corning deal comes with no specific timeline beyond Apple saying that soon every iPhone and Apple watch sold around the world will be built with Kentucky made cover glass. If you heard a little editing snafu at the end there, that's cause I was rushing to get this out to you as soon as possible since I held the show a bit to listen to the keynote from OpenAI. By the way, I recorded this week's bonus episode last night, so the last link in the show notes today is a link to that on YouTube. A conversation with the investor Paul Kadrosky about something I've wanted to talk to you all about for a while now. Could AI data center spending blow up the economy? Check that out now if you want or it will drop in this feed as per usual on Saturday. Talk to you tomorrow.