In a Report and Order, the Commission clarified certain technical details of the degraded throughput methodology used as part of its compatibility analysis of non-geostationary satellite orbit, fixed satellite service (NGSO FSS) licensees. Specifically, the clarification adopts a 3% time-weighted average throughput degradation as a long-term interference protection criterion and a 0.4% absolute increase in link unavailability as a short-term interference protection criterion.
The intent of the changes is to ensure that NGSO FSS licensees authorized in more recent years are compatible with NGSO FSS systems previously licensed by the Commission.
In the same Report and Order, the FCC declined to adopt protection metric modifications or an aggregate limit on interference that were applied in later rounds of NGSO FSS license reviews to licenses reviewed in earlier rounds.
Titled “FDA Digital Health and Artificial Intelligence Glossary—Educational Resource,” the FDA’s glossary includes detailed definitions of more than 50 different terms. Of course, it includes several of the more widely-used terms like “Artificial Intelligence,” “Digital Twin,” and the “Internet of Things.” But the glossary also includes several terms probably known only to those immersed in developing and leveraging the relevant technologies, such as “Convolutional Neural Network,” “Data Drift,” “Explainability,” and “Federated Learning.”
Each term presented in the glossary also includes links to the sources of the terminology, as well as a list of related terms.
The FDA notes that its Digital Health and Artificial Intelligence Glossary is intended only for educational purposes and does not represent FDA policy nor constitute legally enforceable requirements.
The report, titled “Voice Telephone Services: Status as of December 31, 2023,” clearly shows the significant decline in the use of legacy wireline technologies in favor of mobile wireless technologies. Specifically, as of the end of 2023, wireline technologies accounted for just over 18% of the 471 million retail voice telephone service connections in the U.S., with mobile voice subscriptions equaling nearly 82% of total service connections.
Further, the report highlights the growing trend in the shift to mobile wireless technologies. According to the report, during the three years from 2021 through 2023, mobile voice subscriptions increased at a compound annual growth rate (CAGR) of 3.1%, while interconnect VoIP (wired) subscriptions declined at a CAGR of 1.6% and retail switched access (wired) lines declined at a CAGR of 15.7% year.
As our use of artificial intelligence (AI) grows exponentially, researchers and scientists are beginning to sound the alarm about the potential environmental effects linked to AI.
According to a recent article posted to the website of Scientific American magazine, scientists are already estimating that the projected global use of AI by the year 2027 could consume as much electricity as all of the Netherlands. Further, a separate study published in Nature Computational Science predicts that generative AI applications could add between 1.2‑5 million metric tons of electronic waste worldwide by the year 2030.
The Scientific American article provides details on some of the stark consequences associated with the use of generative AI. For example, creating two separate images using AI-generative technology can consume as much energy as charging a smartphone. And a single exchange with ChatGPT can lift the temperature of the physical server generating the exchange enough to require the equivalent of a bottle of water to bring the temperature back to a normal level.
To help alleviate the potential environmental impact linked to the use of AI, the article identifies some potential solutions to help minimize the environmental impact of the use of AI. The list includes more efficient chip and AI algorithm design, more regular maintenance and updating of data servers, and finding more ways to refurbish or reuse obsolete hardware components. Researchers project that implementing these and other strategies could reduce e-waste associated with AI by as much as 86%.