Google's DeepMind has developed an AI that predicts the weather better than any existing system, according to MIT Technology Review.
This is Google's second foray into weather prediction via AI, and it's working out better than ever. This time, the model was trained on 30 years of weather data. Once it was fully trained, it produced predictions better than Ensemble Forecast, the current best forecasting tool, about 97% of the time.
However, meteorologists don't need to worry. According to Scientific American, these models aren't ready to take their jobs. In fact, most meteorologists take data from several models, put it together, add in their specific knowledge of a particular location, and then produce a forecast. The better the models are, the better their forecasts will be.
|
When was the last time you weatherized your home? Click your choice to see results and speak your mind. |
This year 👍 As MIT Technology Review explained, the DeepMind forecast also needs to be interpreted by people who know what they're doing. It doesn't predict particular weather, but instead offers a percentage of a chance that something will happen. So the model might say, "There's a 30% chance of fog in the morning," but it still needs a person to look at actual conditions to determine whether or not that fog will happen.
Additionally, the model is trained on historical data. Since weather is fluctuating around the world as a result of the planet's overheating in recent decades, that historical data may not be enough for the AI to predict what will happen in the future. Professional meteorologists need to look at actual conditions, especially when they don't line up with the model's prediction.
There are concerns about the negative effects of AI on the environment, as outlined by the United Nations' Environment Programme. AI data centers require a lot of power to run and water to keep them cool. There's also a chance that AI could harm animals by aiding poachers and illegal wildlife traders.
While researchers need to do more work to determine the long-term effects, we certainly need to keep the weather pros around either way. Even with AI, we'll need meteorologists to properly interpret, contextualize, and relay the information.
Voting for pro-climate candidates can help ensure that the planet will be considered when future decisions are made about AI. You can also limit your own AI use until we know more and encourage industries to hire creative people rather than depending on technology.
Join our free newsletter for weekly updates on the latest innovations improving our lives and shaping our future, and don't miss this cool list of easy ways to help yourself while helping the planet.
Cool Picks
