AI with Kyle Daily Update 159

Today in AI: Claude Can Now Show You Things

Claude's New Visualizer

Anthropic quietly dropped a new feature this week. Claude can now create interactive visualisations directly in chat. Sliders, charts, animated explainers. All seamlessly in chat.

I had some trouble forcing it to use the new feature. The key trigger is the phrase "show me." As in "show me how compound interest works" or "show me the difference between these two datasets." Other phrases like "draw me" or "create a visualisation" are hit and miss.

Also, if you want to SHARE you visualisations first convert them into an Artefact. Then you can share a public link. No intuitive but that’s the method.

I tested it all morning. Built an interactive compound interest calculator with sliders. Got it to visualise how LLM chunking works with adjustable parameters. Then I ran my actual A/B test data through it - four email welcome sequences, 1,000 people each - and it spat out a proper analysis I could share with my business partner.

The best bit? It works on ALL plans, including free. It's an app-level feature, not tied to a specific model.

Kyle's take: It's useful. It's not world-breaking. People saying this will "destroy" data visualisation businesses are overreacting. As usual.

The "show me" trigger is the thing to remember! Use that to force Claude into visualiser mode.

TikTokers Are Lying About AI Water Usage (I Did the Maths…)

There's a viral trend on TikTok right now. Creators posting "ask me a question - I use less water than ChatGPT." It's everywhere. And it's complete rubbish.

Using a platform like Tiktok (and watching these “ask me a question” videos) uses vastly more water and electricity than AI. That fact is just conveniently ignored!

I sat down and ran the actual numbers using conservative estimates. Per hour of use: ChatGPT (about 30 queries) transfers roughly 180 kilobytes of data. TikTok (60 minutes of video) transfers about 3 gigabytes. That's a massive difference in server load, electricity, and cooling water.

I then used Claude’s new visualiser to create interactive data visualisations:

The water usage people talk about comes from two places - direct cooling of data centres and indirect water used to generate electricity (steam turbines at gas and coal plants). Both scale with how much data you're moving. And video streaming is HEAVY on both.

Even using the most generous numbers for ChatGPT and the most conservative for TikTok, TikTok uses roughly 60x more water per hour…

Same for Instagram, Youtube, Netflix etc. etc. It’s just that people are picking and choosing what to be offended about!

Kyle's take: This one has annoyed me. People on TikTok and Instagram are using "AI bad for environment" as a cheap hook to attract anti-AI viewers. They haven't checked the numbers. They're just repeating something that sounds good. If you're going to criticise AI's environmental impact - fine, there are very legitimate arguments! But don't do it while streaming 4K video for six hours a day. Get your facts together.

The AI Pricing Split Is Creating Two Classes of User

Here's something I've been watching for months that's becoming really obvious now.

People using free ChatGPT get bad results. They don't know to select thinking mode from the dropdown. They use whatever model the auto-selector picks. They hit limits constantly. They conclude AI is overhyped and useless. So they never pay. Downward spiral.

Instagram Reel

People paying for Claude or premium ChatGPT get genuinely good results. They see the value. They're willing to pay more. They get better at using it. Upward spiral.

This split is only going to get wider. In a year or two, the gap between people who invest in learning these tools and people who tried the free version once will be massive. The free version of any AI tool right now is basically a demo from six months ago. It's not representative of what these tools can actually do. If you're serious about using AI in your business, the £20/$20 a month for a Pro plan is the cheapest investment you'll make all year.

Good news: You are already on the right (upward!) spiral by being here reading this right now. But hopefully this helps explain a little about why others may think you’re mad for going all-in on AI.

Want the full unfiltered discussion? Join me Monday for the daily AI news livestream where we dig into the stories and you can ask questions directly.

Until then, keep PROMPTING!

Kyle