There may be an apparent paradox - bordering on seeming like a giant con - in AI's potential to do its bit to green the world while it keeps guzzling enormous amounts of energy in the process.

Actually, though, it's less of a paradox than it's made out to be considering AI will help squeeze more out of every energy source, including fossil fuels , as it helps economies make their transition to RE sources.

Any incremental energy demand, particularly of the kind that increases the carbon footprint, to run the additional computing coming on stream should be offset by less energy-intensive production overall and a quicker switch to alternative sources.

If Microsoft needs to pitch AI to Big Oil while claiming the tech can solve problems such as climate change , it isn't contradicting itself.

AI needs to overcome the cost hurdle to its dispersal that energy companies can help it achieve.

AI will have to overcome three constraints to its widespread application.

Apart from the demand for energy, it will need enormous computing capacity that the semiconductor industry is still unprepared for. The nature of chips will themselves need to be rethought to do justice to universal application of generative AI. Likewise, bots will only deliver if there are enough people who can make them do so. ML can improve only if curricula for humans are transformed to get the best out of the tech.

Each of these involves costs that the tech industry cannot shoulder on its own. So, beneficiaries, much like today's oil companies, will have to pitch in. AI will improve chip design and manufacturing.

However, chip-makers must be on board.

The drug industry stands to benefit immensely from AI, only if it equips medical researchers with some degree of tech-training.

Again, no paradoxes here.

AI, like any technology before it, will have to become commercially viable for it to become a transformative social tool.

Some of the social costs must be paid for while the tech is in development.

In the case of AI, it will be planetary.