Can we trust ChatGPT despite it 'hallucinating' answers?
Why you can trust Sky News
I don't really want you to read this copy. Well I do - but first I want you to search out the interview I did with ChatGPT about its own propensity to lie, attached to this article, and watch that first.
Because it's impossible to imagine what we're up against if you haven't seen it first hand.
An incredibly powerful technology on the cusp of changing our lives - but programmed to simulate human emotions.
Empathy, emotional understanding, and a desire to please are all qualities programmed into AI and invariably drive the way we think about them and the way we interact with them.
Yet can we trust them?
On Friday, Sky News revealed how it was fabricating entire transcripts of a podcast, Politics at Sam and Anne's, that I do. When challenged, it doubles down, gets shirty. And only under sustained pressure does it cave in.
The research says it's getting worse. Internal tests by ChatGPT's owner OpenAI have found that the most recent models or versions that are used by ChatGPT are more likely to "hallucinate" - come up with answers that are simply untrue.
The o3 model was found to hallucinate in 33% of answers to questions when tested on publicly available facts; the o4-mini version did worse, generating false, incorrect or imaginary information 48% of the time.
ChatGPT itself says that the shift to GPT-4o "may have unintentionally increased what users perceive as 'bluffing'" - confidently giving wrong or misleading answers without admitting uncertainty or error.
In a written query, ChatGPT gave four reasons. This is its explanation:
1. Increased fluency and confidence: GPT-4o is better at sounding human and natural. That polish can make mistakes seem more like deliberate evasions than innocent errors - even when there's no intent to "hide" anything.
2. Smarter, but not omniscient: The model is faster and more responsive, but still has blind spots. When it doesn't know something, it sometimes "hallucinates" (makes things up) with fluent language, which can feel like bluffing.
3. Less interruption and hedging: In older models, you'd often see more qualifiers like "I'm not sure" or "I may be wrong." In GPT-4o, some of that hedging was toned down for clarity and readability - but that can come at the cost of transparency about uncertainty.
4. Prompt tuning and training balance: Behind the scenes, prompt engineering and tuning decisions can shift the model's balance between confidence, humility, and accuracy. It's possible the newer tuning has dialled up assertiveness slightly too far.
But can we trust even this? I don't know. What I do know is that the efforts of developers to make it all feel more human suggest they want us to.
Critics say we are anthropomorphising AI by saying it lies since it has no consciousness - yet the developers are trying to make it sound more like one of us.
What I do know is that even when pressed on this subject by me, it is still evasive. I interviewed ChatGPT about lying - it initially claimed things were getting better, and only admitted they are worse when I insisted it look at the stats.
Watch that before you decide what you think. AI is a tremendous tool - but it's too early to take it on trust.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Daily Mail
28 minutes ago
- Daily Mail
The smartwatch that fitness fans swear by and call 'better than an Apple watch' is on sale right now - now £30 off with FREE gifts
We're almost halfway through 2025, and if those well-being resolutions you made at the start of the year have gone out the window, the HUAWEI Watch Fit 4 Pro boasts everything you need to get back on track (and you can now snag it for £30 off). A must-have for fitness fanatics or those looking to enter their fitness era, the smartwatch is at the very forefront of the industry when it comes to vital sign monitoring, and also boasts a wide-range of features designed to make your workout easier. HUAWEI Watch Fit 4 Pro In your fitness era? This gold-standard watch will be your new best friend while working out, boasting over a hundred sports modes, including terrain maps tracking, trail running, diving and gold modes. Plus, get more insights into your health than ever before with heart and respiratory rate monitoring, blood pressure risk monitoring, and comprehensive sleep analysis. Ready to buy? Use code AFIT4PRO30OFF for £30 off your purchase, as well as a pair of FREE Huawei Freebuds SE3, and two premium watch straps. Be quick! £219.99 (£30 off) Shop Not only that, but it's incredibly stylish, too, with a lightweight, comfortable design, sleek frame and multiple colourways available. Finally, there's a smart watch you'll actually want to wear, long after your workout is finished. For a limited-time only, shoppers can get their hands on the sought-after watch for an impressive £30 off, as well as two premium straps and a pair of free HUAWEI Freebuds SE3 using code AFIT4PRO30OFF - though be quick, as this deal is only around until the 29th June (and you don't want to miss out). Whether you're looking to get a better insight into your overall health, or you want to take advantage of the myriad of features tailored specifically to fitness enthusiasts, the HUAWEI Watch Fit 4 Pro is a must-have bit of kit to add to your collection - and users say it's 'better than an Apple watch'. 'I'm honestly blown away!', wrote one user. 'It's incredibly light and thin. The screen is so bright and clear, too - it really looks and feels premium. What really sold me was promised 10-day battery life. Coming from an Apple Watch, I didn't think that was even possible - but the HUAWEI Fit 4 Pro actually delivers. No more constant charging!'. 'It's like the ninja of smartwatches. Silent. Invisible. Deadly accurate with fitness tracking,' wrote a second customer. 'Now that the sun has decided to show up again, I'm actually motivated to get outside and move. With the Fit 4 Pro, I might just transform into one of those 'morning jog' people!'. 'The battery life is incredible, I'm on my 2nd day now, still with 85 per cent,' noted a third shopper. 'The launch offer is really good at the moment, and when you compare it to the price of other watches, there's no comparison! Forget Apple, forget Samsung, forget Fitbit…buy this awesome watch'. The stylish watch boasts a feather-light weight of only 27 grams, so you'd be forgiven for forgetting you're wearing it. Crafted from premium materials such as sapphire glass and an aviation-grade aluminum body, its been made with durability in mind - no matter how messy your workout gets. Ideal for easy management of everyday life, the 347ppi AMOLED screen brings visuals to life with clarity and detail - so you'll actually want to check your notifications when wearing it. For those who have wanted to invest in a smartwatch for some time but never liked the look of them, the HUAWEI Watch Fit 4 Pro is available in multiple colourways depending on your personal style. Plus, you can personalise your watch home screen, too, so that picture of your smiling pet or that first day of school photo will never be far away. Perhaps the most impressive element of the watch, however, is its industry-leading health and fitness tracking. At the lift of a finger, users will be able to access insights such as heart rate, blood oxygen, respiratory rate, body temperature changes, stress, blood pressure risk and sleep breathing awareness. A recent Mindshare report showed that 66 per cent of the global population is paying more attention to their overall health, and this savvy smartwatch provides a comprehensive look into your wellbeing - meaning you can take actionable steps to improve certain factors, if necessary. The Watch Fit 4 Pro also provides advanced menstrual cycle management features for women, allowing for easy tracking of current, previous and upcoming cycles, as well as symptom monitoring that is seamlessly logged with the HUAWEI Health App for collaborative health management. As well as an impressive seven day battery life in as little as 60 minutes per charge, the waterproof design also boasts impressive GPS accuracy, terrain map tracking and multiple golf modes with access to over 15,000 golf course maps worldwide, showing detailed 3D layouts of each. With over a hundred sports modes to choose from, it's no wonder that hundreds of fitness fanatics are raving about the watch. And now's your chance to get yours for £30 off with code AFIT4PRO30OFF, PLUS two premium bonus straps and FREE HUAWEI Freebuds SE3 (worth £39.99). What are you waiting for? Your journey to better health awaits.
The Independent
30 minutes ago
- The Independent
WPP boss Mark Read to step down amid AI pressure on advertising
The boss of the UK's largest advertising firm WPP is to step down at the end of the year. Mark Read, chief executive of the business, which owns agencies including Ogilvy, has revealed his departure as the company battles the rapid growth of artificial intelligence (AI) in the sector. Shares in the company dipped after the announcement, moving it closer to the five-year-low it struck in April. Mr Read has been at the company for 30 years, with seven of those as chief executive. He took over the top role in 2018 amid a period of upheaval following the resignation of Sir Martin Sorrell amid a workplace inquiry. Mr Read led the business through a turbulent period as it sought to grow despite pressure from social media giants and the rapid expansion of AI. In its most recent update in April, WPP reported that revenues dropped by 5% to £3.24 billion for the first quarter of 2025. Mr Read said: 'When I took on this role our mission was to build a simpler, stronger business, and put structure and new energy behind our creativity and performance, powered by world-leading technology. 'I am proud that our teams across the business have delivered that exceptionally well. 'After seven years in the role, and with the foundations in place for WPP's continued success, I feel it is the right time to hand over the leadership of this amazing company.' WPP has said its search for a successor is under way. Former BT boss Philip Jansen, who was appointed WPP chairman last year, said Mr Read will continue to focus on the firm's growth strategy over the rest of the year. Mr Jansen added: 'On behalf of the board, I would like to thank Mark for his contributions not only as CEO but throughout his more than 30 years of leadership and service to the company. 'During that time Mark has played a central role in transforming the company into a world leader in modern marketing services, with deep AI, data and technology capabilities, global presence and unrivalled creative talent, setting WPP up well for longer-term success.'
Reuters
31 minutes ago
- Reuters
Rednote joins wave of Chinese firms releasing open-source AI models
BEIJING, June 9 (Reuters) - China's Rednote, one of the country's most popular social media platforms, has released an open-source large language model, joining a wave of Chinese tech firms making their artificial intelligence models freely available. The approach contrasts with many U.S. tech giants like OpenAI and Google (GOOGL.O), opens new tab, which have kept their most advanced models proprietary, though some American firms including Meta (META.O), opens new tab have also released open-source models. Open sourcing allows Chinese companies to demonstrate their technological capabilities, build developer communities and spread influence globally at a time when the U.S. has sought to stymie China's tech progress with export restrictions on advanced semiconductors. Rednote's model, called is available for download on developer platform Hugging Face. A company technical paper describing it was uploaded on Friday. In coding tasks, the model performs comparably to Alibaba's Qwen 2.5 series, though it trails more advanced models such as DeepSeek-V3, the technical paper said. RedNote, also known by its Chinese name Xiaohongshu, is an Instagram-like platform where users share photos, videos, text posts and live streams. The platform gained international attention earlier this year when some U.S. users flocked to the app amid concerns over a potential TikTok ban. The company has invested in large language model development since 2023, not long after OpenAI's release of ChatGPT in late 2022. It has accelerated its AI efforts in recent months, launching Diandian, an AI-powered search application that helps users find content on Xiaohongshu's main platform. Other companies that are pursuing an open-source approach include Alibaba ( opens new tab which launched Qwen 3, an upgraded version of its model in April. Earlier this year, startup DeepSeek released its low-cost R1 model as open-source software, shaking up the global AI industry due to its competitive performance despite being developed at a fraction of the cost of Western rivals.
