While AI is a sketchy and, let's put it as it is, taboo topic as of late, but it is absolutely a useful and interesting tool in some aspects. In other aspects, its use cases are often debated. It's potential is higher in some cases, in others it's done more harm than good.
There are advancements being made to make AI "eco-friendly" via smaller models, better cooling systems (potable water doesn't get released into the middle of nowhere), and advancements made on chips, which is quite significant to note. By 2030 AI will be reduced significantly enough to have net-zero environmental impacts or, at the very least, no noticable drawbacks towards the environment, as Google, Amazon, and Microsoft all state. Microsoft, Google
To note a specific examples where AI excels, AI has analyzed the data and identified risks of brain tumors within different sexes, AI has detected ovarian cancer early in patients, and has recognized opiod-usage-disorders. UW School of Medicine and Public Health
AI has also made advancements in Sexual Health, according to the National Library of Medicine, which is fully funded and operated by the United States government. AI is used in several core tenets of Sexual and Reproductive Rights, like added support for antenatal (conception until labor), and perinatal (from pregnancy up to a year after birth), prevention and treatment of STIs (Sexually Transmitted Infections), and cervical cancer, contraception and infertility services, etc. National Library of Medicine
AI is also a useful tool for casual users with it's capabilities to summarize large documents and troubleshooting appliances. However, it's advancements in the medical field are particularly impressive.
Google has implemented their AI tool, Gemini, throughout its search. Statistically, Google Search runs approximately 3.5 billion searches each day. This is, by all means, a lot of water usage for simply trying to find an answer for a question on your device. One AI generated image takes 0.3 to 2.0 kWh, the same amount of energy as charging a smartphone. MIT Technology Review, PCMag
A single google search takes ~0.00004 kWh (0.04 Wh) Energy per Search (EpS). A google search with 'AI Overview', uses about ~0.00024 kWh (0.24 Wh). That is significantly more energy usage when compared to the non-augmented result. 8.5 billion searches per day times 0.00029 kWh = 840,000 kWh per day of pure AI energy usage. But lets be generous, and subtract a little for queries the AI simply cannot answer due to it's safety guidelines. Google intentionally disables AI Overviews for queries that are NSFW, or keywords that require a simple and direct answer or redirect. Because of this, Google's AI Overview triggers for only about 20% of all Google searches. DigitalApplied
So with this is mind, it's worth dragging our estimate down to about 408,000 kWh per day.
AI Data centres use up about 1.0 - 2.0 litres of water per kWh of energy they use. Assuming a middleground of 1.5 litres per kWh, 408,000 kWh of energy are used on a daily basis, bringing our water loss estimate to 1,530,000 litres of water per day. OECD, Nature
According to Magazine Coalition, 7 of every 10 searches end without the user being redirected to a single website. CNN has lost between 27% to 38% of it's traffic throughout 2025 due to AI Overview alone. According to Falia, when an AI Overview appears in search results, 8% of readers actually click a link. That's a severe drop in traffic, and that traffic can often be vital for smaller businesses. Falia, Magazine Coalition
Publishers looking to opt-out of the scrubbage are completely wiped out of all searches, leading to even less visitors. The lack of options force publishers for their articles to be scrubbed.
Due to the ongoing surge in AI usage which harms independent publishers, users opposed to AI are searching for a way out of the Google-dominated internet ecosystem, moving to smaller, non-AI oriented search engines instead.
Some parts of AI shine brighter than others. In the medical field, AI has been a lifesaver and a useful tool, able to identify patterns in X-ray scannings, combat cancer, etc. However, it's damage to the industry on the internet — the industry used to research for this very article — has done more harm than good, convenient or not.
Note, my use of em dashes is just correct grammar... I didn't use ChatGpt, I swear !