Forgot your password?
typodupeerror

Submission Summary: 2 pending, 146 declined, 47 accepted (195 total, 24.10% accepted)

Submission + - Performance of a LLM on the reasoning tasks of a physician (science.org)

sinij writes:

In all experiments, the LLM outperformed physician baselines and displayed continued improvement from prior generations of AI clinical decision support. Our study suggests that LLMs have eclipsed most benchmarks of clinical reasoning, motivating the urgent need for prospective trials.

The future of healthcare looking more bleak as there is a push to automate diagnosis. While LLMs are very capable in some areas, notable weakness is relevance realization. That is, humans are by far more capable of developing heuristic models on how to restrict search space and focus on relevant information. Medical diagnosis is one such area. I fear misdiagnosis of rare conditions going to become commonplace as LLMs are widely adopted in the medical space.

Submission + - US government ramps up mass surveillance (theconversation.com) 2

sinij writes:

People have little choice when buying devices, using apps or opening accounts but to agree to lengthy terms that include consent for companies to collect and sell their personal data. This “consent” allows their data to end up in the largely unregulated commercial data market. The government claims it can lawfully purchase this data from data brokers. But in buying your data in bulk on the commercial market, the government is circumventing the Constitution, Supreme Court decisions and federal laws designed to protect your privacy from unwarranted government overreach.

Still nothing to hide?

Submission + - Google Planning an Open Source Platform for Android Auto (caranddriver.com) 1

sinij writes:

Google has announced a new open-source version of Android Auto that the company is calling Android Automotive OS for Software Defined Vehicles, or AAOS SDV. The new system is a more powerful version of Android Auto (AAOS), similar to the new Apple CarPlay Ultra, with the Google version taking things a step further.

This is Step 1 on the way to automotive enshittification, where unskippable in-dash ads are played before startup and driver-facing camera is used to enforce ad-watching.

Submission + - A Private Company Wants to Block the Sun (theatlantic.com) 1

sinij writes:

Stardust sold geoengineering to investors. Now it needs to sell it to the public.

More like extract public funds doing something extremely reckless. We have no idea what intervention like that would do to our ability to grow food.

Submission + - Big Tech deserves its Big Tobacco moment (marketwatch.com) 1

sinij writes:

Landmark verdicts shatter the Section 230 shield, turning ‘addictive’ product design into a legal thicket for Meta, Alphabet and others.

The fact that social media is designed to be addictive is now court-tested fact.

Submission + - State Farm insurance uses drones and AI image analysis to drop coverage (nypost.com)

sinij writes:

Linda Bennett, who has lived in her Santa Ana home since 1993, said she was stunned when she received a notice warning that her roof needed to be replaced or she risked losing her coverage. The project is estimated to cost roughly $20000.

Euphemistically calling it "aerial roof assessments", insurance providers using it to perform automated inspections without human review. However, these are not always accurate, as various conditions (morning dew, etc.) can lead to false positives.

Submission + - Honda Cancels All Three EVs That It Planned to Build in the US (caranddriver.com)

sinij writes:

The automaker was blunt in its announcement of the changing plans, citing American tariff policies and the unpredictable nature surrounding American EV incentives and fossil fuel regulations. In its release marking the announcement, Honda made it clear that it expected to incur further financial losses over the long term if it went through with launching the cars.


Submission + - Canadians advocating for nationalized AI (schneier.com)

sinij writes:

Imagine AI embedded into health care, triaging radiology scans, flagging early cancer risks and assisting doctors with paperwork. Imagine an AI tutor trained on provincial curriculums, giving personalized coaching. Imagine systems that analyze job vacancies and sectoral and wage trends, then automatically match job seekers to government programs. Imagine using AI to optimize transit schedules, energy grids and zoning analysis. Imagine court processes, corporate decisions and customer service all sped up by AI.

To me, this sounds dystopian, because I can also imagine AI declining your permits, renewal of license, or medication due to misalignment or 'greater good' reasons.

Slashdot Top Deals

Diplomacy is the art of saying "nice doggy" until you can find a rock.

Working...