Unleashing the Power of Gemini, CoPilot, and chatGPT4o in AI

Gemini, CoPilot, and chatGPT4o my! - AI: Actionable Insights #8

Welcome to this edition of AI: Actionable Insights – grab a drink and let's get into it.

There’s a challenge I have putting each edition of this newsletter together – the relentless pace of change in AI - for the last 3 editions, between starting the research and writing, something significant has changed meaning a major re-write. I delayed this edition to make sure I included the big 3 who all had announcements planned.

In the last edition we talked about how we’re close to tipping point of mass adoption (Everyone, everywhere and all very soon) this took another step forward with all the news from Google IO, the Microsoft CoPilot + PC announcement, and the OpenAI chatGPT4o release.

All of them showcased real time interactive voice updates that are multi-modal and closer to a real-life Jarvis – the rumours are these capabilities from OpenAI are coming to Siri, Google have been very clear that their assistant will soon include Gemini, and CoPilot are building AI right into the laptop.

OpenAI launched their latest model chatGPT4o to ALL users, free and paid.

This is a much more capable and faster model than was previously available to free users and includes Memory, Data Analysis and Vision in the free version for the first time.

There’s a good set of explainer and demo videos on the Hello chatGPT4o page.

As a paid user you get 5x the usage limits and significantly enhanced Data Analysis tools – I haven’t seen many people talking about this second feature but it looks much more useful for generating charts and analysing data, live interactions with created tables as well as it now connecting to Google Drive and OneDrive.

You can now interact and edit tables and charts created by chatGPT - read more on Enhanced Data Analysis here.

Note: this is “rolling out in the coming weeks” – I don’t have it yet, but I’m looking forward to experimenting with it as soon as I can!

Google IO Updates

Google IO started with a chaotic performance from Marc Rebillet showing off their Generative MusicFX DJ… if you’re not familiar with him, he’s certainly a colourful choice to launch a show with: Google I/O 2024: Pre-Show

In this demo, Google shows Gemini in two contexts: one with the prototype running on a Google Pixel phone and another on a “prototype glasses device”... interesting inclusion in itself as they've not announced any new Google Glass products...

Gemini Google IO

Another big update is an increase in the context size for Gemini Ultra to 2 million tokens – this allows the equivalent of a bookshelf worth of data to be uploaded and analysed, significantly larger than any other model, I’m sure we’ll see some phenomenal use-cases opened up by this expansion.

They also announced a new model Gemini Flash, which is a significantly faster and cheaper model.

Side note: there are now by my count 6* products with the Gemini name: Gemini Ultra, Gemini Advanced, Gemini Pro, Gemini Nano, Gemini Flash, Gemini for Workspace, and if you include Gems (similar to custom GPTs) that’s 7.

Microsoft CoPilot + PC Event

Last night's Microsoft event saw them launch their new evolution in computing – PCs with AI built in at every level, on the machine and via the internet with CoPilot + PC:

There’s an odd mismatch with CoPilot – because it's from Microsoft, corporate businesses are happier to use it – but I, and others, still feel it’s notably less capable than other models (see Getting real about CoPilot) – that said I’ve noticed some improvements in the last week so I’m still open minded.

Microsoft CoPilot