The Download
#003
Dear Reader,
Greetings again from an incredibly muggy Valencia. Yesterday and today I’m attending the VDS 2024 Tech Conference at the City of Arts and Sciences, so this update will be coming in a little late this week as I spend time attending, distilling and reflecting the experience.
Valencia’s City of Arts and Science. Even better when it’s sunny.
In this week’s edition: Updates on Perplexity, Claude’s Anthropic, reflections on creating my “digital twin” and more.
Let’s get to it.
📈 Levelling up my LinkedIn
For the next four weeks I’ll be taking part in the Remote Work Europe ‘Level Up Your LinkedIn Challenge’, so if you’re following me on that platform, you’ll notice an uptick in posts from now until mid-November.
I’m using Notion to keep myself on track with what I want to talk about as well as which formats I want to adopt.
I don’t normally write in Notion, but I find it’s a great database app for tracking actions and information retrieval.
👩🏽💻Super Prompts for Solopreneurs
Welcome new readers who downloaded the Super Prompts for Solopreneurs Pack earlier this week. For those of you who have already subscribed, you can get the pack here.
🔍 Perplexity follow-up
Last week I wrote about how I was using Perplexity.ai as my default search engine, and I’m still going strong with it and enjoying all that it has to offer.
While I’m in balancing getting my AI consultancy off the ground, I’m also still looking for remote contract roles. Perplexity was extremely useful this week in helping me draft specific cover letters and specific versions of my CV that pulled from my existing experience, then framing them so they were more specifically tailored towards the roles. More on that next week. In the meantime you can check out my short explainer on getting started with Perplexity here.
🤖 Anthropic’s Claude can use your computer
Anthropic unveiled significant updates to its Claude AI models this week, including a groundbreaking new feature called “Computer Use”, allowing it to perceive and interact with computer interfaces, enabling it to use a wide range of tools and software like a human would. Put simply, it can use your computer.
There is tremendous scope here beyond automating drudge work and repetitive procedures, as demoed above. Think of the opportunities for disadvantaged users, such as those with accessibility issues.
While this is still in beta for API use only, there are obvious security implications that need to be taken into consideration. I can’t wait to try it out.
Apple take note: this is where Siri should be heading.
♊︎ Digital twin reflection
Well, I paid my money and I took my chance with Synthesia, as I wrote about last week. My attempts to create a short form video with an AI clone of myself didn’t go well.
While Synthesia certainly looks better than other popular rival HeyGen (in my opinion), it still just wasn’t close enough for me to be happy calling it a digital twin.
Creating a script and plugging it into the product was easy enough, and the results came back faster than the originally scoped 10 minutes. It was clear, however, that I was going to have to spend more time tweaking and waiting to possibly get closer to results.
But in that amount of time, I could just shoot a video on my phone and already be on my way towards editing it. 🤷♂️
I believe there is still a place for creating static, scripted content that doesn’t require a lot of dynamic emotion, perhaps for training videos. While that’s a major selling point for Synthesia, it does also lean more into marketing the creation of more realistic models than the competition is capable of. But I don’t think it’s there just yet.
I still use ElevenLabs.io for voice cloning, and I think that is still a good platform, but conversely this week we have seen the release of F5 TTS, a text-to-speech synthesis program that can be run locally on your own computer. Watch this space.
🎨 Gen-AI in art
This demo from Adobe last week got me seriously excited for practical creative uses of generative AI. You don’t need to be a graphic designer, or an Adobe user to get the nuance behind why this is important, and it’s all around a the computer vision occlusion problem.
Unlike humans, who can move around and change perspective, most computer vision systems rely on static images or video feeds with a fixed viewpoint. This means they cannot simply look around obstacles to see what’s behind them. The demonstration above shows that generative AI is now at a point where it can create what it previously could not infer based on what it could “see”.
That’s why there’s a cheer when Turntable shows the other obscured legs of the horse.
As this was a part of Adobe’s “Sneaks” at their MAX conference, there’s no guarantee that this will make it to a final product, but it was an incredible demonstration of creative, generative AI in action.
🔗 Quick links
- The Breakfasteur. Watch as this self-described “Doctor Mom” takes her child on a fun exploration of the human body, through the lens of a surgeon. With Play-Doh.
- Apple’s Business Connect Program. Apple announced some new updates to their Business Connect Program, including ways to include your business logo on messages and phone calls.
👋🏻 Wrapping up
VDS took up most of my time this week but it has certainly left me with a lot to think about. But I am looking forward to getting stuck back in with some actual tool development next week.
AI-curious? Don’t hesitate to get in touch.
Have a great rest of your week!
Jim
Made with ❤️ in Valencia by Jim Christian. For feedback, please reach out to [email protected]. For custom GPTs, prompt libraries, general AI consulting and development, please visit Informatic AI.
