As of 2024, Microsoft no longer has active data centers underwater. If you recall, the tech giant had its experimental research and development called Project Natick . It was first proposed in 2013 to determine how an underwater farm server could help reduce cooling costs while promoting environmental sustainability. The first prototype was deployed during the initial phase in 2015, followed by a second phase in which hundreds of servers were submerged 117 feet below. By 2020, Microsoft declared that underwater data centers were reliable and practical. But it seems Microsoft is already ready to move on. Noelle Walsh, the corporate vice president at Microsoft’s Cloud Operations + Innovation division, confirmed that there are no plans to build more underwater data centers. She said her team worked on the project, yielded results, and gained knowledge about operating servers below sea level. The company will leverage the knowledge gained from the experiment and apply it elsewhere.
If you have wondered where the concept of Artificial Intelligence originated, the answer lies in the many computer scientists who have been closely studying and attempting to imitate the human brain for years.
These computer scientists have tried to copy the structure of our brain through neural networks with amplified processing skills to come up with what we now know simply as “AI.” However, the more technology advances and the potential of AI is harnessed frequently, the more energy is consumed.
Ironically, this is when scientists return to nature where a Swiss start-up company recently launched what they dub as a “biocomputer,” one that can attach to functioning, pulsing brain cells. The company also states that it uses substantial amount of energy than traditional, bit-based computers.
FinalSpark’s (the said startup company) online platform harnesses spherical clusters of laboratory-produced human brain cells known as organoids instead of simply merging biological ideas into computing. This research in organoids is significant because it is studied at a period when artificial neural networks have gained popularity.
There is a total of 16 organoids kept within four arrangements that are then connected to eight electrodes each, as well as a microfluidics system that provides water and nutrients for the cells.
This is known as wetware computing, where researchers’ abilities are put to use to culture organoids in the lab. It’s a recent technology that lets scientists study what are technically mini replicas of various organs.
- Low income may contribute to faster brain aging — study
- LIST: Massive asteroids that impacted Earth
- Scientists say coin tosses are not really 50/50
FinalSpark’s comment on their developed technology
According to the company, the brain-machine interface system bioprocessors that they are continuously developing takes up a million times less power than traditional digital processors.
While there are no specific numbers yet for factors such as energy usage or processing power, the research team comments that teaching just one large language model like the GPT-3 already requires 10 gigawatt hours, equivalent to 6,000 times the energy that a single European citizen consumes in a year.
On the other hand, our brain’s 86 billion neurons consume only a fraction of that same energy, a total of just 0.3 kilowatt hours per day.
Latest reports on technology also mentions the rise of the Artificial Intelligence industry and how frequent use of AI has the capacity to consume 3.5 percent of the world’s electricity by 2030. The IT industry alone has already contributed 2 percent of global CO2 emissions.
FinalSpark as not the only company returning to nature for assistance
While the Swiss corporation has developed their biocomputers, they are not entirely the first company to try harnessing mother nature in attempt to connect probes to biological systems or even program neural networks so they work on particular input-output functions upon command.
In fact, researchers from the United States also built a bioprocessor just last year that allows a computer hardware to attach to brain organoids. The system was quick to recognize speech patterns with such a bioprocessor.
At current the system serves as a way for researchers and computer scientists to perform lengthy experiments on brain organoids, just like those that have already come before it. Of course, the objective of having a new, energy-efficient computing approaches is never far from these experiments.
With the continuous experiments and development, FinalSpark can confidently say that researchers can now connect to its system remotely, with the sustainability of the mini-brains lasting up to 100 days (their electrical activity is monitored 24/7.)
A final note from the company states that, this year, the system is available for research purposes and is free for researchers to utilize to work towards the end goal.
The start-up company hopes that in the future, they can extend the capabilities of their platform to manage a wider range of experimental protocols significant and relating to wetware computing. This would include injecting molecules and drugs into organoids for testing.
We look forward to more updates on FinalSpark’s continuous efforts to achieve a perfected biocomputer and make use of mother nature balanced with the rise in technology.
Source: ScienceAlert
This article, Swiss company develops low energy “Biocomputer” using 16 human mini-brains, was originally published at NoypiGeeks | Philippines Technology News, Reviews and How to's.
Comments
Post a Comment