Written by Agency Inc
Like it or not, we have all fallen for AI.
It’s everywhere. Writing novels, personalising our social feeds, creating self-driving cars, and automating factories. Every day it goes to work, spotting patterns in our lives, predicting our next moves, understanding our words.
It can be an incredible tool. But while some of us might be smitten with the power of AI, our planet isn’t.
In fact, some experts are warning that the advance of artificial intelligence could lead to an 80% increase in our global carbon emissions.
Worryingly, these fears seem justified.
A recent study by the Blair Institute found that training a large language model like GPT-3 can emit 552 tons of carbon dioxide equivalent. With the increasing demand for AI and its energy-intensive nature, these numbers are set to soar. In fact, Goldman Sachs estimates a 160% increase in data centre power demand due to AI between 2023 and 2030.
As we pointed out in a previous post, a single ChatGPT query requires nearly 15 times as much electricity to process as a Google search. And the BBC Artificial Human podcast “How green is my AI?” revealed an AI-generated image consumes the equivalent energy of about half a smartphone battery.
So, what can we do? How can marketers use AI for good – but limit its impact on the planet?
Here are some suggestions:
- Always ask about AI policies when you’re choosing partners. If your agency uses AI constantly and unnecessarily in the background, you’re not just being robbed of the human expertise you’re paying for. Your marketing is also impacting the planet – more than you know, and without your control.
- Assess AI applications against each other – and don’t just keep adding new platforms or use cases. It’s all about return on energy: what is truly worth the planetary cost? Our next few pointers give some guidance on this.
- Take inspiration from real success stories before trialling AI in your organisation. What has genuinely worked – not just within marketing, but beyond? This article is a good place to start. Many successful examples are more about streamlining workflows and quickly summarising information – not competing high-value tasks like strategic or creative thinking.
- Always ask: as a customer or prospect, what would matter to me? Whether it’s an add-on for your marketing automation software or a customer support agent, there are plenty of possible applications for AI. But some of them leave the customer far from impressed. Generative AI, for example, can help customers by providing faster information access, delivering very simple communications, or speeding up documentation. But if you need the customer to feel wowed or understood? Or to believe in your depth of expertise? AI may work against your goals – however much time you think you’re saving.
- Focus on back-end applications first. AI is superb at data crunching. In fact, new tools can analyse customer profiles and make sure individuals get served the content they’re more likely to enjoy.¹ It’s not about writing or creating content – just personalizing the experience. This sort of back-end application could make the biggest difference for your marketing department initially.
Beyond the personal steps we can take to control our AI use, we also need support from the tech industry. Because right now, it’s hard to know exactly how much energy AI is using.
We need more information so we can make smart choices. If we know how much energy different AI tools use, we can pick the ones that are better for the planet. It’s like knowing the miles per gallon of a car. Or rather, kWh/mile (since many of us are going electric).
The more we know, the better decisions we can make.
What else do you need to know, to make better decisions on AI? Get in touch to tell us.
We make the complicated simple.