Even I Can't Keep Up (And This Is My Job)
Even I Can't Keep Up (And This Is My Job)
I spend a lot of my working life reading about AI. I've got automations that pull news into one place, summarise it for me, and turn it into content.
This week, I fell behind.
Not by much. Just a few days of catching up. But when I finally sat down to sort through the headlines, something jumped out.
I'd organised them alphabetically by company. And there, under 'O', was OpenAI. Again. And again. And again.
What OpenAI Launched in September Alone:
- Projects for free users
- Conversation branching
- £36 million grant fund
- Startup accelerator programme
- Major coding tool upgrades
- Teen safety features
- UK infrastructure deal
- NVIDIA partnership worth billions
- Team collaboration tools
- Parental controls
- In-app shopping with instant checkout
- Complete video generation platform (Sora 2)
- Social media app to rival TikTok
- Content moderation system
That's 15+ major product launches. In one month. From one company.
Is This Normal?
Maybe. Perhaps this happened with previous technology waves and I wasn't paying attention. I was probably at the pub or watching football.
But it doesn't feel normal. It feels relentless.
The Real Problem
I take a daily bath in this stuff. I get paid to understand it. And I find it overwhelming at times.
So how do people on the edge feel?
How do you keep up when you've got an actual job to do? When you're trying to run a business, manage a team, hit targets?
No wonder so many people feel under-equipped. No wonder they avoid getting started. The pace of change isn't inspiring. It's paralysing.
What Actually Helps
Not reading every announcement. Not trying to master every tool. Not waiting until you feel ready.
What helps is understanding the handful of capabilities that matter for your work. Learning to spot the difference between a genuine breakthrough and a feature update. Knowing which tools solve real problems versus which ones are just impressive demos.
That's what education looks like in this moment. Not comprehensive understanding. Practical competence.
Free AI Cook-a-Long Sessions
I deliver AI Cook-a-Longs to leadership teams and departments across the UK. Sixty minutes of hands-on practice where teams learn the core AI capabilities that actually matter for their work.
I'm now running a free weekly version. Same content. Same hands-on approach. Limited to small groups so everyone gets proper support.
The next free session is Thursday @ 10am.
Latest AI News
The Brain Drain Problem
MIT study: AI makes your brain work less, and the effects linger
MIT researchers found that using ChatGPT doesn't just make your brain work less hard during a task - it continues to reduce brain activity in future work even without AI.
The Trust Problem
ChatGPT secretly switches models when you get emotional
OpenAI quietly routes emotional or sensitive prompts to stricter, more restrictive models without telling users.
The Coding Wars Continue
Claude Sonnet 4.5 can now code for 30+ hours straight
Anthropic's latest model tackles complex coding tasks for over 30 hours without losing focus.
DeepSeek slashes API prices by up to 75%
Chinese AI company DeepSeek released V3.2-Exp with prices 50-75% cheaper than before.
Google's Jules coding agent gets command line access
Google added terminal interface and API access to its AI coding agent Jules.
OpenAI beats humans at world's top programming contest
An OpenAI system solved all 12 problems at the International Collegiate Programming Contest World Finals 2025.
The Global AI Divide
Google warns EU is losing the AI race to China
Google's President of Global Affairs says EU AI adoption sits at just 14% whilst China's companies report 83% adoption.
OpenAI's models are steeped in Indian caste bias
MIT Technology Review found that GPT-5 reproduces harmful caste stereotypes 76% of the time.
AI and Wikipedia are destroying vulnerable languages
Machine-translated Wikipedia articles in small languages are riddled with errors.
New Tools and Benchmarks
Samsung launches real-world enterprise AI benchmark
Samsung's TRUEBench tests AI models on actual workplace tasks across 12 languages.
The Philosophy Corner
Physicist: True AGI needs its own story
Physicist David Deutsch argues you can't test for general intelligence like you test software.
The Implementation Reality
Why 95% of AI pilots fail: It's the operations, not the tech
A Lucid Software study found that 61% of workers believe their organisation's AI strategy is poorly aligned with operational capabilities.