A report from Apple has cast significant doubts on the efficacy of reasoning models, going as far as to suggest that when a problem is too complex, they simply give up.
AI reasoning models have emerged in the past year as a beacon of hope for large language models (LLMs), with AI developers such as OpenAI, Google, and Anthropic selling them as the go-to solution for solving the most complex business problems.
However, a new research paper by Apple has cast significant doubts on the efficacy of reasoning models, going as far as to suggest that when a problem is too complex, they simply give up. What's going on here? And does it mean reasoning models are fundamentally flawed?
In this episode, Rory Bathgate speaks to ITPro's news and analysis editor Ross Kelly to explain some of the report's key findings and what it means for the future of AI development.
Highlights
"It sounds a bit obvious, but at the same time, the way that this has been, framed by a lot of providers, makes it seem like they're the be-all and end-all. But once you reach a level of complexity, yeah, you start to encounter some serious problems."
"What's interesting for me, right off the bat, is, what Apple is showing is that that's actually the opposite of what these are good at, that they are maybe more performant, maybe they're slightly better at solving complex problems up to a point, but that if you actually try and give them really complex problems, they are completely ineffective."
"Apple is claiming that OpenAI hasn't made a model that can think, to which my reaction was, did you think OpenAI had made a model that can think? "
- Subscribe to The IT Pro Podcast on Apple Podcasts
- Subscribe to The IT Pro Podcast on Spotify
- Subscribe to the IT Pro newsletter
- Join us on LinkedIn
-
IT management and security in the modern age: Moving away from fragmentation towards seamless successSponsored Content Proper endpoint, user, and business management demands a unified approach
-
Data (Use and Access) Act comes into forcenews Organizations will be required to have an effective data protection complaints procedure and fulfil new requirements for online services that children are likely to use
-
‘A complete accuracy collapse’: Apple throws cold water on the potential of AI reasoning – and it's a huge blow for the likes of OpenAI, Google, and AnthropicNews Apple published a research paper on the effectiveness of AI 'reasoning' models - and it seriously rains on the parade of the world’s most prominent developers.
-
Dell Technologies World 2025: All in on AIITPro Podcast From the cloud to the edge, enterprise architecture for AI must be carefully considered
-
SAS Innovate 2025: Pursuing pragmatic AIITPro Podcast As CISA delivered an eleventh-hour reprieve for the CVE database, AWS reportedly began to pause some data center leases
-
Putting small language models under the microscopeITPro Podcast The benefits of small language models are undeniable – but they're no silver bullet
-
Google Cloud Next 2025: Targeting easy AIITPro Podcast Throughout its annual event, Google Cloud has emphasized the importance of simple AI adoption for enterprises and flexibility across deployment
-
Is the IT service desk dying out?ITPro Podcast Employees could soon use chatbots to resolve common IT issues rather than wait for tickets to be resolved
-
Harnessing unstructured data to fuel AISponsored Podcast Knowing how your data is stored can make all the difference to your AI success
-
How effective are AI agents?ITPro Podcast AI agents have been sold with the promise of fully automating tasks without the need for step-by-step instructions
