ChatGPT boss compares new AI to inventing nukes and reveals he's terrified of super-smart bot
CHAT'S TERRIFYING ChatGPT boss compares new AI to inventing nukes and reveals he's terrified of super-smart bot
Click to share on X/Twitter (Opens in new window)
Click to share on Facebook (Opens in new window)
CHATGPT mastermind Sam Altman has opened up about his AI fears saying the tech has left him feeling "useless".
The OpenAI chief likened the feeling of spearheading AI to that of scientists behind the Manhattan Project, a WWII research and development programme devised to create the first nuclear weapons.
Sign up for Scottish Sun
newsletter
Sign up
2
Altman says he felt 'useless' compared to the AI's recent enhancement, GPT-5
Credit: Getty
A new version of the powerful ChatGPT system, GPT-5, is set to be unleashed soon with enhanced capabilities in understanding and processing like never before.
Altman, 40, said it "feels very fast" when testing the update recently.
After receiving a question he didn't quite understand via email he put it in GPT-5 which "answered it perfectly".
"I really kind of sat back in my chair and I was just like, oh man, 'here it is' moment," he told the Theo Von podcast.
"I felt useless relative to the AI, in this thing that I felt like I should have been able to do and I couldn't, it was really hard, but the AI just did it like that.
"It was a weird feeling."
The OpenAI boss went onto talk about moments in the history of science where you have a group of scientists look at their creation and say "what have we done?".
"Maybe the most iconic example is thinking about the scientists working on the Manhattan Project in 1945, sitting there and watching the Trinity test...
"It was a completely new, not human scale kind of power and everyone knew it was going to reshape the world.
"And I do think people working on AI have that feeling in a very deep way. You know, we just don't know."
ChatGPT CEO Sam Altman admits feeling anxious over how powerful AI will change our lives as Congress warns of 'new industrial revolution'
Altman also laughed as he was likened to a "Charming Terminator".
It's not the first time Altman has been frank about his fears around AI.
He's previously admitted that "if this technology goes wrong, it can go quite wrong".
What was the Manhattan Project?
Here's what you need to know... The Manhattan Project was a US WWII research and development project intended to create the first nuclear weapons
It was led by the US but was also supported by the UK and Canada
It started because of fears that the Nazi's had created nuclear weapons during WWII
The project started in 1939 and continued into the late 1940s
Most of the project was based at a facility in New Mexico
It created the nuclear bombs which were dropped over Hiroshima and Nagasaki in 1945
This remains the only use of nuclear weapons in armed conflict and killed between 129,000 and 226,000 people
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Scottish Sun
6 hours ago
- Scottish Sun
My clever new fridge saves me hundreds on food AND electricity bills – it even has a cool Ring doorbell perk
Click to share on X/Twitter (Opens in new window) Click to share on Facebook (Opens in new window) FRIDGES and freezers are getting in on the smart stuff these days - and they can save you a heap of money in the long-run too. The latest flash model from Samsung has done just that for me, both on food and my electricity bill. Sign up for Scottish Sun newsletter Sign up 5 The Sun's Jamie Harris gives a hi-tech fridge a go Credit: Jamie Harris / The Sun 5 Answering Ring doorbell calls is my favourite trick Credit: Jamie Harris / The Sun 5 It will help you save loads on food shopping too by cutting down on waste Credit: Jamie Harris / The Sun There are a number of mind-boggling techy features your everyday fridge can't do too. Some are better as party tricks to show off to friends but the majority of them have turned out to be super handy in my house. The exact Samsung model I'm trying is the 621L Bespoke AI Series 9 Side-by-Side Fridge Freezer with AI Home (RS90F64EDTEU). So what's so special about it? Read on in my review below to find out. Samsung Side-by-Side Fridge Freezer review The fridge door has a little display which is your portal to a number of apps. More and more fridge freezers are adopting these sorts of screens - and at one point they were quite big. This one is small and unimposing, allowing you to access things like Spotify and YouTube. It's touch screen and voice powered, so you can essentially do away with having an Alexa or similar digital assistant in your kitchen, especially as the speaker can go pretty loud. The voice assistant is Samsung's own, Bixby, so you'll have to say "Hey Bixby" to summon anything. Samsung reveals genius way to help you save money on food using AI in your FRIDGE For the most part, I just use the touch screen - but there's one special trick that's very much Bixby's. And that's the ability to open the doors for you. At first, I dismissed this as a bit of a gimmick (and was what I referred to as a "party trick" earlier). However, truth be told, there have been times when I've been cooking, I have my hands full with items and suddenly this feature has come in handy. The handles do have sensors on them too that automatically opens the door for you when you simply hover your hand over it - I also thought was just plain lazy initially but I've grown to make that the norm all the time now. A minor annoyance to mention, sometimes when I opened the fridge door my elbow passed the sensor for the freezer door causing it to open too. 5 The screen is a perfect size Credit: Jamie Harris / The Sun Saving money on food and more What I really care about is saving money and I'm shamefully aware that my food waste could be a whole lot better. Sometimes you forget, other times you have no idea what to do with that one ingredient. Well, Samsung's SmartThings helps you cut food waste, thereby reducing how much I spend on shopping. You can log items via the app on your phone - which is available on both Android and iOS, not just Samsung's handsets - or tap them in on your fridge screen, along with their use-by dates. While it's very manual, I've resorted to treating it as my shopping list app as opposed to just using my phone's notes feature as I did before. The fridge will also suggest recipes based on the ingredients logged for some easy ideas, crafted by the likes of Jamie Oliver and more. There are other ways the fridge helps you save money too. For instance, it learns your usage patterns so that energy is optimised at the right times. You can even tell it when you're on holiday so it'll go into an energy saving mode. The screen will tell you exactly how much electricity the fridge has used too (last month it was just shy of £10). And as there is a water and ice dispenser, the tech tells you when it's time to change the filter too. Inside When it comes to the primary function, keeping your food chilled and frozen, the inside is perfectly ordinary. In the fridge there are five shelves, as well as two fruit and veg compartments at the bottom. And there are four shelves with two storage draws along the bottom of the freezer. Another thing to note is that it's pretty quiet as a fridge freezer goes, despite being a pretty big appliance. 5 A look inside the fridge and freezer Credit: Jamie Harris / The Sun Ring doorbell trick My favourite trick of all, arguably, is the simplest. And that's integration with Ring doorbell. SmartThings is compatible with hundreds of smart home brands, not just devices made by Samsung, which is a rare treat. Once you've linked your Ring doorbell into the SmartThings app it means the fridge screen will alert you when someone is at the front door. You can even answer the call there and then, so another reason not to have another smart hub sitting around your kitchen. Samsung Side-by-Side Fridge Freezer: conclusion At £2,239, the 621L Bespoke AI Series 9 Side-by-Side Fridge Freezer with AI Home doesn't come cheap. But it's worth noting you get five year warranty for parts and labour, as well as 20 year warranty on digital inverter compressor. You can also knock up to £100 off if you recycle an old appliance with Samsung when making the purchase. But for £2,239 you get a premium, slick and feature-packed appliance that should save you hundreds - if not thousands - in the long run with those food and energy management features. Rating: 4/5
Telegraph
a day ago
- Telegraph
Private ChatGPT conversations leak onto Google search results
Hundreds of ChatGPT conversations have been exposed in Google search results – revealing alleged discussions of insider trading, potential fraud and doctors' questions. Researchers uncovered more than 500 conversations that were available via the search engine, as well as thousands more that had been archived elsewhere on the web. This week, ChatGPT's developer OpenAI blocked web searches from collecting conversations, saying people may be accidentally sharing private information. It came after a series of compromising conversations were discovered in Google's search results by Henk van Ess, an online researcher. The chats involved discussion of insider trading plans and admissions of fraud, as well as attempts to plan a cyber attack against Hamas. Other conversations appeared to involve doctors and lawyers asking for advice on their work. Another involved one user seeking to manipulate the chatbot into generating inappropriate images of children. Conversations with chatbots such as ChatGPT are typically private, but users can share records of the conversations on social media or messaging apps by creating a hyperlink. In some cases, OpenAI allowed shared chats to become discoverable by search engines, meaning they would show up in Google and other sites' search results. The company said on Thursday that it had disabled the feature. Dane Stucky, OpenAI's chief information security officer, said: 'Ultimately we think this feature introduced too many opportunities for folks to accidentally share things they didn't intend to, so we're removing the option. We're also working to remove indexed content from the relevant search engines.' Cheating schoolchildren However, Mr van Ess said a further 110,000 conversations could be found through the Wayback Machine, a tool from the technology non-profit the Internet Archive which stores snapshots of the web. The discussions included an Italian lawyer looking for advice on removing native communities for an energy project in the Amazon, and an Egyptian dissident criticising the country's president. Conversations frequently involved schoolchildren demanding that the chatbot write essays on their behalf. The conversations are anonymous but in many cases users include personal information, either through chats or documents they have uploaded. Sam Altman, OpenAI's chief executive, recently said that there was no legal confidentiality related to conversations with ChatGPT. 'People talk about the most personal s--- in their lives to ChatGPT. People use it – young people, especially –as a therapist, a life coach. 'If you talk to a therapist or a lawyer or a doctor about those problems, there's legal privilege for it ... We haven't figured that out yet for when you talk to ChatGPT.'
Geeky Gadgets
a day ago
- Geeky Gadgets
How OpenAI Agents SDK Makes AI Observability Effortless
What if you could peer into the inner workings of your AI systems with the same clarity as a mechanic diagnosing an engine? In the fast-evolving world of artificial intelligence, understanding how your agents operate isn't just a luxury—it's a necessity. Yet, many developers and engineers struggle with the lack of transparency in AI workflows, leaving them guessing about performance bottlenecks or inefficiencies. Enter the OpenAI Agents SDK, a tool designed to transform AI observability. With its built-in tracing capabilities, this SDK offers a window into your agents' behavior, providing actionable insights that can transform how you debug, optimize, and scale your AI-driven applications. In this exploration, James Briggs uncovers how the OpenAI Agents SDK enables you to monitor and refine your AI systems with precision. From tracking response times and token usage to customizing workflows for large-scale projects, the SDK's features are tailored to meet the demands of modern AI development. But it's not just about metrics—it's about the stories those metrics tell. Whether you're troubleshooting a sluggish agent or fine-tuning a complex workflow, the tools at your disposal promise to make the process intuitive and impactful. By the end, you'll see why mastering AI observability isn't just a technical advantage—it's a strategic one. OpenAI SDK Tracing Overview Built-in Tracing: A Comprehensive Monitoring Solution The tracing functionality within the OpenAI Agents SDK offers a robust framework for observing and analyzing agent workflows. It captures critical data points such as response times, token usage, and tool outputs, providing actionable insights into your agents' efficiency. Setting up tracing is straightforward and requires only an API key and basic configuration through the OpenAI dashboard. Once enabled, the system automatically records key metrics, allowing you to focus on analyzing the data rather than managing the setup process. This functionality is particularly useful for identifying inefficiencies or bottlenecks in your workflows. For example, if an agent's response time is consistently slow, tracing data can help pinpoint whether the issue lies in the agent's logic, the tools it uses, or external dependencies. Managing Access and Permissions for Enhanced Security To maintain security and privacy, tracing data is accessible only to organization owners by default. If you are working within a team, you can adjust access permissions through the OpenAI dashboard to share logs with other authorized engineers. This controlled access ensures that sensitive tracing data remains secure while allowing collaboration. By carefully managing permissions, you can maintain the integrity of your project's observability while fostering teamwork. How OpenAI Agents SDK Enhances AI Debugging and Scaling Watch this video on YouTube. Check out more relevant guides from our extensive collection on AI Observability that you might find useful. Customizing Traces to Fit Your Workflow The OpenAI Agents SDK offers significant flexibility in tailoring tracing workflows to meet specific project requirements. Using the `trace` function, you can define custom workflows and group IDs, making it easier to organize and analyze traces. Additionally, metadata can be attached to traces, enhancing filtering and search capabilities. For instance, tagging traces with project names, workflow stages, or specific objectives can streamline navigation and improve trace management. This customization is particularly beneficial for large-scale projects involving multiple agents or workflows. By organizing traces effectively, you can quickly locate relevant data and focus on optimizing key areas of your application. Streamlined Search and Filtering for Large Data Sets Navigating large volumes of tracing data can be challenging, especially in complex projects. The OpenAI Agents SDK addresses this issue with advanced search and filtering tools. These features allow you to quickly locate traces based on workflow names, group IDs, or metadata. By allowing efficient data navigation, these tools save time and help you focus on the most relevant information. For example, if you are troubleshooting a specific agent's performance, you can filter traces by the agent's name or associated metadata. This targeted approach simplifies the debugging process and enhances overall efficiency. Debugging and Optimizing Performance Tracing is an invaluable tool for both debugging and performance optimization. By examining metrics such as response times, token usage, and tool outputs, you can identify inefficiencies and areas for improvement. For example, if an agent consistently generates slow responses, tracing data can reveal whether the issue stems from the agent's logic, the tools it employs, or external dependencies like the OpenAI web search tool. This level of insight enables you to make informed adjustments, improving the overall performance and reliability of your AI-driven applications. Whether you are addressing specific issues or conducting routine performance evaluations, tracing provides the data needed to refine your workflows effectively. Practical Applications of Tracing The tracing capabilities of the OpenAI Agents SDK are versatile and applicable across various scenarios. Some practical applications include: Monitoring an agent's adherence to custom instructions. Evaluating the performance of tools like the OpenAI web search tool. Analyzing token usage to optimize cost efficiency. These use cases highlight how tracing can provide actionable insights, allowing you to refine and optimize your agents' behavior in real-world applications. Limitations to Be Aware Of While the tracing features offer significant advantages, it is important to recognize their limitations to set realistic expectations: Tracing is not supported in Google Colab by default and requires a local environment for proper functionality. The OpenAI web search tool may exhibit slower response times and variable output quality, which can impact performance analysis. Understanding these constraints allows you to plan your workflows more effectively and mitigate potential challenges. Key Benefits of Tracing with OpenAI Agents SDK The tracing features in the OpenAI Agents SDK provide several benefits that simplify debugging and performance analysis: Detailed monitoring of agent workflows with minimal configuration. Customizable traces to track specific parameters or workflows. Advanced search and filtering tools for efficient data navigation. Actionable insights to optimize agent performance and reliability. These advantages make tracing an essential tool for developers and engineers working on AI-driven applications. Maximizing AI Observability with Tracing The OpenAI Agents SDK's built-in tracing capabilities serve as a powerful resource for improving AI observability. By allowing detailed monitoring, offering customization options, and providing advanced search and filtering tools, the SDK equips you with the tools needed to gain deeper insights into your agents' performance. While there are some limitations, such as the need for a local environment, the benefits far outweigh these challenges. Whether you are debugging issues, optimizing workflows, or analyzing performance metrics, tracing enables you to enhance your AI projects effectively and efficiently. Media Credit: James Briggs Filed Under: AI Latest Geeky Gadgets Deals Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.
