💥 A Chinese AI has just made all our hard drives obsolete

At the current rate, the production of digital data far exceeds that of storage capacity, which requires making choices, such as not keeping certain information, or degrading the quality of recorded data.

To address this acceleration in storage needs, researchers from Fujian Normal University in China have developed an original process that simultaneously exploits several properties of light to record information in three dimensions. This method allows for much more data to be housed in a volume equivalent to traditional processes, thus offering a serious avenue to meet demand.


Most current systems, such as hard drives, record data on a flat surface. The new technique, however, relies on the principle of holography to inscribe information within the thickness of a material. By superimposing several patterns within the same volume, this approach significantly improves storage density.

The idea of holographic storage is not new, but the innovation here lies in the combined use of three properties of light to form 'pages' of optical information.

Light is not just a simple luminous flux. It notably possesses a property called polarization, which describes the direction in which its wave vibrates. The researchers have managed to make this polarization a stable and reliable information channel, in addition to the intensity and phase properties already used.

To achieve this, they applied a three-dimensional modulation technique that allows a single optical device to control these three parameters simultaneously and in a coordinated manner. The writing is reliable and fast enough for most needs.

However, reading this multidimensional data has until now faced a significant technical difficulty. Standard sensors can only directly measure the intensity of light in a single reading, leaving other information inaccessible.

To overcome this hurdle, the scientists used an artificial neural network. After training on diffraction images, it became capable of reconstructing all three dimensions from simple intensity measurements. It analyzes the small subtleties in the patterns to deduce the missing information about phase and polarization in the recording.

Once the concept was validated by experiment, the researchers observed that this combined approach effectively increased the amount of information carried by a single holographic page. Simultaneous decoding by artificial intelligence also greatly speeds up reading, since it avoids multi-step measurements. This alliance between optics and AI thus enables storage that is both dense, with reading that is reliable and fast.

For now, this technology remains at the experimental stage. Several improvements are required before any potential deployment. The scientists plan to increase the resolution of the encoded data and strengthen the robustness of storage materials sensitive to polarization.

The ultimate goal is to combine this method with other multiplexing techniques to record multiple data pages in parallel, and thus further push the capacity limits.