Osegi
Science·2 min read

First black hole image confirmed by independent AI analysis

An Australian team re-ran the raw Event Horizon Telescope data with machine learning and got almost the same M87* ring, without telling the model what to expect.

The first image of a black hole, a bright orange ring of hot plasma around a dark central shadow in galaxy Messier 87.
Image: اليوم السابع

Is the first picture of a black hole real? An independent team says yes. Australian researchers rebuilt the image from scratch using AI and ended up with a picture almost identical to the one the world saw in 2019.

The original image came from the Event Horizon Telescope (EHT), a worldwide network of radio telescopes. It shows the supermassive black hole at the centre of the galaxy Messier 87, about 55 million light years away. The black hole, known as M87*, weighs roughly 6.5 billion times as much as the Sun. Some later studies had questioned how accurate the final shape was, since combining data from telescopes spread across the planet can pick up technical errors and distortion from Earth's atmosphere.

Samuel Lai of Australia's CSIRO led a small team that built a new technique called GenDIReCT, based on machine learning. Lai compared imaging such a distant object to photographing a doughnut on the Moon from your back garden. The model was not told in advance what shape to expect, yet it recovered the bright ring around the dark shadow. Lai calls it the first truly independent check of the EHT image.

How did they photograph a black hole 55 million light years away?

The EHT links radio telescopes around the world so they act like a single dish the size of Earth. The method is called Very Long Baseline Interferometry. It gives enormous sharpness, but huge distances, weather differences and equipment errors can distort the data. The Australian team's fix was to combine measurements from groups of three telescopes rather than pairs, which cut calibration errors while keeping the real shape of the object.

Why it matters for Einstein

Einstein's general relativity, published in 1915, says mass bends spacetime and that gravity comes from that bending. It also predicts black holes and a ring of hot, bright plasma around a central shadow. The study, published in Nature, measures the ring at about 42.2 microarcseconds across, give or take 1.5, which fits earlier mass estimates.

Laurent Loinard, director of the EHT collaboration, said independent verification is essential for a discovery of this size. The researchers also say the same methods could help with medical imaging and earthquake mapping.

Apps worth a look

SUMO: the number puzzle

Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.

Tags:black holem87event horizon telescopemachine learninggeneral relativitycsiro

More from this day

See all →

More in: Science

See all →