#Weeknotes 112 (21 Feb) — More consequences and benefits of AI and its potential to make us better humans.

.

Work wise:

AI has been on my mind a lot lately, and I’m sure I’m not alone. This week, our team gathered to discuss how AI can be applied to our work in meaningful ways, ways that truly add value to our clients. It was an engaging session, starting with an open discussion about our attitudes toward AI and what it means for our work as design consultants.

Most of us shared a similar perspective: a cautious optimism. We recognise that AI is here to stay, and rather than resist it, we need to understand it. As UX designers, we know that solving real-world problems starts with deeply understanding them. Yet, in the AI gold rush, many organisations seem to have lost sight of this. The race to integrate AI into everything has led to a flood of shiny new AI-powered features, often launched without clear use cases. The expectation? That users will somehow figure out how to make them useful.

This approach is backward. It shifts the burden onto users to justify technology’s existence, rather than starting with what people actually need. AI is just another tool, not a solution in itself. Throwing AI at problems without first understanding them is not only ineffective but potentially harmful.

Beyond its usability challenges, AI also has a significant, often invisible, environmental cost.

ChatGPT explaining to me the energy use of a single prompt in a way that a child can understand. I love how clearly it explains the concept and I can see myself leaning on these tools to explain things to my child.

Since I don’t know the numbers and am simple-minded, it felt appropriate and relevant to use AI to put this into perspective.

Prompt: What’s the equivalent energy consumption for the use of ChatGPT across the world on an average day?

ChatGPT Answer:

The exact energy consumption of ChatGPT globally depends on many factors, including server efficiency, total user interactions, and data center energy sources. But let’s break it down with an estimate.

Step 1: Energy per prompt

A single ChatGPT prompt is estimated to use around 2 to 10 watt-seconds (joules). This is roughly equal to powering a small LED light bulb (10W) for 0.2 to 1 second.

Step 2: Number of prompts per day

ChatGPT gets over 200 million prompts per day (a rough estimate based on OpenAI’s user base).

Step 3: Total energy per day

If each prompt uses around 5 watt-seconds (on average), then:

200 million prompts × 5 watt-seconds = 1 trillion watt-seconds (1 terajoule) per day

Step 4: Comparing to everyday energy use

• Households: This is roughly the daily electricity use of 10,000 US homes (each using ~100 kWh/day).
• Light bulbs: This could power 10 million 10W LED bulbs for 3 hours.
• Cars: It’s about the energy in 8,000 gallons of gasoline.

Takeaway

ChatGPT’s total daily energy use isn’t massive compared to global electricity consumption, but it’s significant. If usage grows, optimizing efficiency will be key to reducing its environmental impact. 🌍⚡

I love the way it breaks down energy use in a way that it’s easy to understand. That aside, the thought that, on an average day, ChatGPT use across the world can equate to powering 10,000 homes, is a sobering reality.

And yet, despite knowing this, like so many of us, I struggle to resist the ease and convenience AI offers. Because, at the end of the day, I’m human. And willpower alone isn’t always enough. And my curiosity gets the best of me, like how I just turned that explainer into a video using an AI video maker:

Within minutes, I was able to get AI to help create a video about AI’s environmental impact in a way that a child can understand. This ‘power’ is hard to resist.

This is why responsibility can’t rest solely on individuals. Organisations and governments need to play a larger role in shaping behaviours that are better for both humanity and the planet. Thoughtful policies and intentional product and service design should guide us toward better choices, and make doing the right thing effortless.

For example, what if every AI chatbot displayed a real-time energy consumption indicator? What if friction were baked into environmentally harmful actions while positive behaviours were made seamless? Subtle design nudges could help us become more conscious of our choices, rather than leaving us to navigate the ethical and environmental costs on our own.

So the question is: How can we design a world where being good is easier than (often unknowingly) being harmful? How do we make the right choices feel natural, rather than a test of willpower?

Because if we want a future where AI works for us, we need to design it with care, before it quietly takes more than it gives.

Life wise:

While I have reservations about AI seeping into every aspect of our lives, along with the immediate and long-term consequences, I can’t deny that I’ve benefited from its use.

The internet revolutionised knowledge, making the entirety of human history and information instantly searchable (thanks to Google). Memorisation became less essential. Facts I once struggled to retain in school, such as key historical figures, periods of innovation and historic events, are now accessible in seconds, no recall required.

Now, AI is replacing another cognitive function: sense-making. Just as it can refine a messy draft of writing into a structured, coherent piece, it can also bring clarity to something even more elusive: our own emotions. Our inner worlds are often messy, complex, and difficult to process. But AI, in some ways, is becoming an interface for our emotions and moods, helping us organise the messiness of being human.

Earlier this week, I had an imposter syndrome moment. On a whim, I turned to ChatGPT, wondering if it could help. And to my surprise, it did. After just a two-minute exchange, I felt noticeably more at ease, reassured, even. It was incredible.

As someone who believes therapy can benefit everyone but acknowledges its inaccessibility for most (myself included), I found this use of AI surprisingly valuable. It offered an “objective” perspective that helped me untangle my thoughts, escape the spiral of self-blame, and pinpoint small actions that could lead to positive reframing. AI may not match the depth of a human therapist, but it’s striking how well it can process and clarify the everyday struggles that come with being human.

Beyond that, AI can adopt different personas — stepping into roles across various professions, offering just the right kind of “ally” when we need it most. This adaptability has the potential to boost confidence, affirmation, and resilience in ways we’re only beginning to explore.

Of course, as powerful as AI is, it remains a tool — one that can be used for good or harm. The responsibility falls on us, both as users and creators, to determine how we wield it.

I’m curious to see how we can shape AI not just to be more intelligent, but to help us be better humans — to create a kinder, more empathetic world.

Things I came across:

A super relevant and powerful quote from the latest DD newsletter:

We have created a Star Wars civilisation, with Stone Age emotions, medieval institutions, and godlike technology.

– Edward O. Wilson

Photo of the week:

Thanks to Apple’s built in accessibility features, we were able to create a toddler friendly interface on the iPad for our 3-year-old. He can watch his favourite shows on CBeebies, and call family members when they aren’t around. This is both child and senior friendly!

Disclaimer: I use AI to revise my original writing, not create content. The stories are my own.