Telecomunicații Știri: 31 august 2025 - 17 septembrie 2025

October 18, 2025, 5:46 PM EDT. At the AAO 2025 meeting, UCLA's Alfredo A. Sadun argues AI is not replacing humans; it is surpassing us by far in ophthalmology. He cites two examples: Google's AI reading fundus images to infer patient gender with about 97% accuracy, far above human performance (the eye paths must be past puberty and pathology-free); and rapid AI detection of papilledema in the ER, which can trigger necessary brain imaging. A retrospective study showed AI edging out two neuropathologists in distinguishing true papilledema from normal discs, mainly due to speed. Yet neuro-ophthalmology remains in short supply, posing practical hurdles like pupil dilation, plus broader social, political, and legal issues that must be worked out.

October 18, 2025, 5:44 PM EDT. At AAO 2025, AI is framed not as a replacer but as a surpasser in certain tasks. UCLA's Alfredo A. Sadun cites two examples: in fundus imaging, Google's AI can determine patient gender from the image about 97% of the time-far beyond typical human performance (puberty and pathology caveats apply). In the ER, AI can detect papilledema quickly, outperforming two expert neuro-ophthalmologists on speed and narrowing the gap in diagnostic accuracy. With a growing shortage of specialists, AI could help ensure patients aren't missed when brain conditions loom. But practical hurdles remain, including who dilates the pupil and broader social, political, and legal issues that must be resolved.

October 18, 2025, 5:42 PM EDT. Researchers at the University of Otago, in collaboration with Google and the Chinese Academy of Sciences, have demonstrated centimetre-level positioning on consumer smartwatches by using multi-GNSS signals (GPS, Galileo, BeiDou, QZSS) and carrier-phase techniques. In four hours of stationary testing, the devices achieved an 8 cm positional scatter, far surpassing prior smartwatch GPS accuracy. The work, led by Robert Odolinski and powered by the Google GnssLogger app, shows that combining signals from multiple systems enables survey-grade precision on wearable hardware. If extended to moving scenarios, this could transform fitness tracking, safety features, health insights, and professional applications, though indoor and urban-canyon performance remains to be proven.

October 18, 2025, 5:40 PM EDT. Researchers at the University of Otago, in collaboration with Google's Android Context group and the Chinese Academy of Sciences, have unlocked centimetre-level precision for smartwatch positioning using multi-GNSS and carrier-phase techniques. In four hours of stationary testing, the watches achieved a repeatability with about an 8 cm error margin. Led by Robert Odolinski, the team used the Google GnssLogger app and combined carrier-phase signals from GPS, Galileo, BeiDou, and QZSS to deliver near-constant, survey-grade accuracy on consumer wearables. This could redefine what wearables can do: more accurate fitness metrics, safer outdoor navigation, enhanced health insights, and new professional uses. The next hurdle is maintaining precision while moving, indoors or in urban canyons, and pushing next-gen GNSS hardware into mass-market devices.

October 18, 2025, 5:34 PM EDT. AI is reshaping business as firms race to harness data centers, software and automation. The article spotlights Palantir Technologies as a core AI enabler expanding from government work to strong private-sector demand. Palantir's commercial revenue growth, $2.3 billion in contract-value bookings, and a 128% net-dollar retention signal accelerating momentum and margins. Though shares trade at a rich multiple, analysts like Dan Ives see Palantir approaching a $1 trillion market cap in coming years. The piece argues that AI adoption creates a virtuous cycle: a company using Palantir gains operating efficiency, prompting more firms to invest in AI platforms. Long term, Palantir's ontology-based digital twin and high-margin software could sustain earnings and free cash flow as AI expands across industries.