Himanshu Gupta on BBC: AI for climate funding, energy demand, and responsible AI
BBC
In a BBC interview, Himanshu Gupta argues that AI built for climate solutions is not receiving funding at the scale of the climate problem, while also discussing how machine learning can help balance renewable-heavy grids and how Climate AI weighs the emissions trade-off of running AI models.
What he said
Is AI for climate getting enough attention and investment?
No. Gupta says funding for AI that is contextual for climate solutions is not close to the scale of the problem.
“No, because the amount of funding required to develop AI which is more contextual for climate solutions is nowhere compared to the scale of the problem.”Watch at 00:24
How does he frame the funding gap?
He contrasts broad AI investment with the much smaller share going toward AI solutions for climate change.
“But the percentage allocated for AI solutions to climate change would be in the orders of $20, $30, $40 million. So that's nothing as compared to the scale of the problem.”Watch at 00:50
What trade-off does he raise about AI priorities?
Gupta says chatbot development is important, but argues that AI priorities should be judged against what matters for humanity.
“So the decision we need to make is like, okay, to what extent do we want to keep on developing AI for developing chatbots? Those solutions are also important. But at the same time, have to have the right perspective of what is more important for the humanity.”Watch at 01:00
Where can AI help in energy systems?
He points to renewable-heavy grids, where variable demand and variable supply create a balancing problem.
“But renewables are very variable, and so is the demand, right? So now we have a problem of balancing demand and supply.”Watch at 01:39
What role do machine learning models play there?
Gupta describes models that understand demand, supply, forecasts, and storage capabilities to optimize renewable dispatch into the grid.
“So that's where many machine learning models are coming up that basically understand the demand patterns from the consumer side or industrial side. And then they also understand the supply patterns based on forecast of solar, wind, and then the storage capabilities and use that to optimize dispatch of renewables into the grid.”Watch at 01:45
How does he address AI's own emissions trade-off?
He says Climate AI weighs human impact against environmental impact, including emissions from its AI models, through a cost-benefit analysis.
“So we do this cost benefit analysis and we measure our impact through like how many farmers lives you've done in this case would be 12 million already. So that compared to the emissions from our AI models, we say like, okay, we are, it's worth the trade-off.”Watch at 02:49
Key takeaways
- AI for climate solutions is not funded at the scale of the climate problem.
- Gupta contrasts broad AI spending with a much smaller allocation for AI solutions to climate change.
- He says chatbot development matters, while asking what is more important for humanity.
- Variable renewable supply and variable demand create a grid-balancing problem.
- Machine learning can help understand demand, supply, forecasts, and storage to optimize renewable dispatch.
- Climate AI weighs the emissions trade-off of its AI models through a cost-benefit analysis.