Why AI is making you stupid, OpenAI's (and Google's) game-changing new releases, and Apple drops the (creative) ball

Thought of the week: AI is already making us lazy

OpenAI should be applauded (and possibly feared) for the impact it has had and will have on the world. Within less than 18 months, it has managed to upend the way we generate content and how we seek answers. However, it’s the seeking of answers that is problematic. As annoyed as I am at Google for polluting Search with ads and search engine optimised content, at least it didn’t do your ‘homework’ for you.

Generative AI, on the other hand, is all about simple, fast and easily digestible answers; i.e. minimising what psychologists call cognitive load. I wrote about this a while ago, and referenced Daniel Kahneman’s excellent work on system 1 thinking (slow and heavy) and system 2 thinking (fast and impulsive). ChatGPT is all about satisfying the human need for ‘fast-food thinking’; generating quick answers that bypass our critical faculties. Essentially, these systems promise to do our thinking for us.

The implications of this are profound, especially with humanised, conversational interfaces becoming the norm. People will become increasingly comfortable with (and eventually dependent on) AI-generated advice.

Take dating, for example. Whitney Wolfe Herd, the founder of Bumble, the female-first dating app, says that in the near future “your dating concierge could go and date for you with other dating concierges … then you don’t have to talk to 600 people. It will go scan all of San Francisco for you and say, ‘These are the three people you really ought to meet’. That’s the power of AI when harnessed the right way.”

She goes on to say, “You could share your insecurities [with the dating concierge]. 'I just came out of a breakup. I have commitment issues.’ And it could help you train yourself into a better way of thinking about yourself. And then it could give you productive tips for communicating with other people.”

If people are prepared to entrust one of the most important decisions of their lives to an AI, then clearly they’ll have no issues outsourcing their system 2 thinking for things like who to vote for. An even more worrying scenario arises if we extend this cognitive outsourcing to how we’re governed and policed. AI in government is the shiny new technological panacea that, following eGov and blockchain, promises to let governments do more with less – especially by making those oh-so tricky system 2 decisions like how to provide effective public services and protect national borders.

The next time you reach for ChatGPT to write an important email, pause, put pen to paper and think instead.

OpenAI and Google up their AI game

The advantage of writing a weekly newsletter is that you almost always have plenty to write about. This week was no exception. OpenAI cemented its lead at the top of the AI stack by rolling out what it called GPT-4o (or GPT-4 Omni). It has lots of stuff under the hood and it's not yet generally available, but the five key things you need to know are:

1. Everyone will soon be talking to ChatGPT rather than typing. Sam Altman has said on many occasions that he was inspired by the film 'Her', and the flirtatious voice used on the demo was clearly OpenAI’s attempt to get us all closer to replacing our friends with a virtual Scarlett Johansson. I’m a huge fan of conversational AI and this update means that you can converse far more naturally (including by interrupting ChatGPT’s often long-winded responses) and quickly following the update.

1. OpenAI is spending a lot of money on making AI friendlier. By humanising ChatGPT, we are more likely to trust it (and outsource our thinking to it!).

2.  The AI will be on your desktop. This means it can be permanently looking at what you’re doing, getting the full context, and soon it will be able control your apps (and again, do your thinking within them).

3. Multimodal is the word. This was always there on the pro version, but now (or soon) everyone will be able to point to something with a webcam/phone and ChatGPT will use that visual information to answer your question. Oh yeah, it can now watch a video and use that as well!

4. OpenAI is very good at PR (but terrible at naming its products). Yes it was nauseating at launch, but it was incredibly slick and impeccably timed to steal Google’s thunder from its massive IO update a few days later.…and that was just OpenAI’s Spring Update.

…and that was just OpenAI’s Spring Update.

Google goes all in on AI

Clearly pissed that OpenAI had managed to upstage them with Omni, Google mentioned ‘AI’ 120+ times during its I/O keynote. Key takeaways:

1. Google will do your googling for you and may come at a cost.

From ‘AI Overviews’ to automatic categorisation, Google is bringing AI to practically every part of the search process. A seismic shift is that Google is now reorienting around answering rather than link-giving, which will massively impact content creators. This shift not only threatens the livelihoods of many web publishers/companies who rely on Google traffic, but also challenges the diversity and accessibility of information on the internet.

1. AI agents are here. Like OpenAI, Google is releasing a new AI sidebar in Gmail. Users can ask Gemini to help them organise receipts from their emails and even put them in a Drive folder, or extract the information from the receipts and put them into a spreadsheet. Gemini can also offer to automate the workflow for use in the future. As I mentioned last week, it is only a matter of time before Google uses all your emails to ‘profile you better’. During the I/O event, the company demonstrated how Gemini could help you return a pair of shoes by locating a receipt in your Gmail, filling out the return form for you, and even scheduling a package pickup. Or, it could help update your address across all the different services that you use after you move. The company says that the agents work under your supervision but are able to reason, plan and think multiple steps ahead.

2. Massive context window. Gemini 1.5 Pro, Google's latest generative AI model, now has the capacity to process around 1.4 million words, two hours of video or 22 hours of audio, which means much improved performance in tasks such as code generation, logical reasoning and complex human conversations.

3. A Sora competitor. Veo can render one-minute videos from a single sentence and Google is trying to win the PR battle by courting many a film creative. If you’re a film-maker, reach out to Google; it’s desperate for good PR.

4. ‘Glassholes’ are coming back - Like OpenAI, Google updated its models to be fully multimodal and revealed something called Project Astra, which included a really cool clip and AI-powered glasses. One my key predictions for 2024 was AI-powered glasses. Interesting to see Google following Meta’s lead in quietly reintroducing them.

Apple drop the (creative) ball, big time

Anyone who knows me knows that I don’t belong to the cult of Apple, so there was a little bit of Schadenfreude when the 'cooler than you' company beloved of the cool kids the world over dropped a supremely uncool ad for their latest iPad last week.

The video centres on a huge waste compactor machine crushing those annoyingly old-school things called musical instruments and vinyl record players, compressing them into the new iPad. The subtext is clear: You don’t need to be a real musician or an artist anymore because the new iPad will do all the creative heavy lifting for you.

The sense that Apple just can’t get it right these days was compounded by their announcement that they may be replacing the detested Siri with an OpenAI voice assistant on Apple devices. It may not yet happen (Apple were also apparently talking to Google about the same thing last year) but just the fact that they exploring this tie-up is a sign that they are falling behind in the AI race.

More robots being abused tested

It’s incredible that you can buy this robot for the price of a family car. The pace of development and the drop in price is dizzying and shows that we’re on track for cheap and ubiquitous I, Robot-type humanoid robots within a few years. Though I do worry about the guy beating up the robot in the video. Seems pretty reckless given the demo of the bot crushing walnuts :-). Let’s hope he remembers to wipe its memory.

What we’re reading this week

• Creatives rising up: writers and publishers in Singapore reject a government plan to train AI on their work

• An astrophysicist claims that AI is the reason we can’t detect aliens

• How ChatGPT will replace ‘senior employees’

• How a Stanford breakthrough may spell an end to clunky VR headsets

• AI will hit the labour market like a ‘tsunami,’ IMF chief warns

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.