The profound implications of 'thinking' AI, Why soon AI will make us all behave, and a new AI frontier in video game development
Thought of the week: the profound implications of ‘thinking’ AI
As predicted, OpenAI released a new series of ‘thinking’ models that far surpass the current state of the art. Rather than just predicting what you want to hear, these models can reason through complex tasks and solve harder problems than previous models in science, coding and maths.
The awkwardly named ChatGPT o1 model can now successfully answer PhD-level physics, biology and chemistry problems, and officially has an IQ of 120!
I’m always suspicious of benchmarks, so I tried my own test. Though I dumped MS Word years ago for Google Docs, I miss a shortcut to change the capitalisation of words. So my test was, could ChatGPT’s new o1 model help me add a little “Toggle Case” button to the menu in Google Doc?
I duly asked the bot and OMG did it deliver. A simple non-technical prompt served up my own custom Google Docs menu. See the video below for my prompt.
Ok, so what I ‘created’ was very simple and procedural, but let’s take a step back here. A non-coder just instructed an OpenAI bot to customise a Google product, with a paragraph of text, which it did successfully, with no errors, in under a minute!
It’s a safe prediction that the vast majority of coding and, by extension, anything digital and possibly analytical will, by the end of 2025, be just post-AI edited.
AI is now at a point where it thinks better, writes better and creates better than all but the best in the field. With AI tools like ChatGPT having doubled their IQ in the space of 18 months and, given the level of spending in the AI arms race, it’s only a matter of (at best) another 12 months or (at worst) a few years before we reach superhuman intelligence. We are all in for one hell of a ride!
Why you need to be using Notebook LLM
Google’s NotebookLM has finally been released to non-US users. Notebook is a research hub/assistant to which you can upload your texts/documents. It recently added an awesome feature which allows users to create a podcast with two AI-generated speakers who literally discuss and walk you through any topic you’re interested in.
This means you can, for example, upload a lesson plan or topic list, and the AI will create an incredibly realistic podcast around this topic. Use cases would be:
an author uploading their book to produce a podcast of it
a teacher creating a podcast for a class topic
a student uploading their course notes to reinforce a lecture
Below is a technical pitch that we sent for a UN project – turned into a podcast.
AI-powered surveillance will ensure you behave, 24/7
Larry Ellison, the co-founder of Oracle, recently shared a frightening vision of the future during a company financial meeting, as reported by Business Insider. Ellison envisions a world where AI-powered surveillance plays a central role in our daily lives. In this future scenario, AI systems would continuously monitor both law enforcement and civilians through an extensive network of cameras and drones.
The goal? To ensure we all behave and do as we're told. Ellison believes that by analysing footage from various sources, including security cameras and police body cams, AI could automatically detect and report criminal activities. This would create a society where, in Ellison's words, "we are constantly recording and reporting everything that's going on".
This Black Mirror-esque dystopia has been riffed on countless times, but the fact that a head of one of the world's biggest tech firms says this sh*t like he's discussing the weather is a sign that this is a destination we cannot avoid.
AI can now create gameplay and will very soon create entire games
The Chinese continue to aggressively push AI as a great accelerator of any digital process. Gaming, with its billion-dollar revenues, but also billion-dollar costs, is an obvious target. Hence, it’s unsurprising that the race is on to produce games cheaply and quickly. Tencent, the biggest name in Chinese-made video games, is investing massively in creating an AI to generate games from simple text prompts
The above scene was created with a simple prompt:
“A person in a trench coat and hat walks along a riverbank, approaching wooden houses on a misty morning. In this atmospheric sequence from the action-adventure game Red Dead Redemption 2, Arthur Morgan is depicted walking along a serene riverbank, highlighted by his distinctive wide-brimmed white hat and dark trench coat. The environment features a tranquil riverside setting bathed in golden sunlight with mist lingering over distant forests. The camera tailing Arthur captures steady shots that subtly reveal more of the lush greenery and rustic buildings emerging on the left bank as he proceeds forward, reflecting the game's characteristic blend of exploration and attention to scenic detail.”
Combining AI scene development with AI coding will mean games can be created in weeks rather than years.
What we’re reading this week
Inside Google’s seven-year mission to give AI a robot body.
A Canadian study showed that an AI early warning system reduced unexpected patient deaths by 26%, monitoring vital signs and alerting staff to intervene earlier.
AI pioneers called for international oversight to address potential catastrophic risks from rapidly advancing AI technology, warning it could soon surpass human control.
Tools we’re playing with
Google's NotebookLM is your personalised AI research assistant. What’s cool is that there's an ‘Audio Overview’ feature that turns your uploaded documents, PDFs or pasted content into two-way podcast-style discussions between two AI hosts.
That's all for this week. Subscribe for the latest innovations and developments with AI.
So you don't miss a thing, follow our Instagram and X pages for more creative content and insights into our work and what we do.