Pradeep Goel

Founder & CEO 30+ years experience in public health, finance, technology & insurance as CEO, COO, CIO and CTO & warrior for user rights.

[Pradeep Goel @_PradeepGoel]

Full article
https://t.co/WDJFhHsJ9Q

[Pradeep Goel @_PradeepGoel]

Researchers at the University of Florida have developed an AI tool called AIDD that can distinguish between Alzheimer’s disease and dementia with Lewy bodies with near-perfect accuracy using specialized MRI scans.

This is significant because the two conditions are often confused despite requiring different treatment approaches. Some studies estimate that up to half of dementia with Lewy bodies patients are initially misdiagnosed, creating the risk of ineffective or even harmful interventions.

[Pradeep Goel @_PradeepGoel]

What makes this work interesting is that the AI is helping detect patterns that are difficult for humans to observe consistently across large volumes of imaging data. This is where AI may create some of its most meaningful impact in healthcare.

Many medical conditions already generate enormous amounts of data through imaging, lab tests, clinical notes, and patient histories. The challenge is often not collecting more information, but identifying the patterns that matter early enough to influence care.

Tools like AIDD point to a future where AI helps surface those patterns with greater precision, giving clinicians stronger evidence for diagnosis and treatment decisions.

[Pradeep Goel @_PradeepGoel]

When intelligence becomes free, verification becomes expensive. The overlooked economic shift, as generation costs approach zero, trust costs approach infinity.

Every AI system creates two parallel markets - one for outputs, another for verification. The latter grows faster.
We're building infinite intelligence with finite verification capacity.

[Pradeep Goel @_PradeepGoel]

AI-managed treasuries are already making allocation decisions without human intervention. This is autonomy.

Agent-to-agent commerce will create financial flows invisible to human oversight. The financial system is being built in machine-readable languages. Autonomous capital will create value.

[Pradeep Goel @_PradeepGoel]

@MarioNawfal The closer AI gets to accelerating science, the harder it becomes to separate between innovation and security policy.

[Pradeep Goel @_PradeepGoel]

@TheAtlantic Assigning responsibility to the tool is often a convenient way to avoid asking who designed it, deployed it, or benefited from it.

[Pradeep Goel @_PradeepGoel]

@yaroslavwr_ Institutions usually adopt infrastructure upgrades before they adopt structural change. That's why this path is more likely.

[Pradeep Goel @_PradeepGoel]

@AlexanderKalian Biology has always been limited more by experimentation than ideas. AI helps with the ideas part, but the experiments still take time.

[Pradeep Goel @_PradeepGoel]

@deredleritt3r The technology may arrive gradually, the challenge is that institutions tend to adapt much slower than technology does.

[Pradeep Goel @_PradeepGoel]

@tszzl Compute is not scarce in absolute terms, it's scarce relative to demand. That's what keeps pushing the industry toward efficiency.

[Pradeep Goel @_PradeepGoel]

@DaveShapi Productivity growth without broadly distributed ownership can create very different outcomes than productivity growth alone.

[Pradeep Goel @_PradeepGoel]

@GaryMarcus The first prediction is the one I find most plausible. Every generation of computing eventually gets pushed toward better efficiency

[Pradeep Goel @_PradeepGoel]

@DeryaTR_ Every major technology changed what people could do. AI may change who gets to do it

[Pradeep Goel @_PradeepGoel]

@craigweiss The question is whether AI forces institutions to become less bureaucratic, or whether institutions force AI to behave like bureaucracy.

[Pradeep Goel @_PradeepGoel]

The web was built for humans. AI agents consume information differently. What does an agent-native website look like? Structured data, API-first, minimal cognitive overhead.

What is SEO when the reader is not human? Signal-to-noise optimization, not keyword optimization.

The internet's evolution is agent-centric. The internet properties of 2035 will be designed primarily for machine consumption, with human access as secondary consideration.

[Pradeep Goel @_PradeepGoel]

https://t.co/xPcwQC8ra0

[Pradeep Goel @_PradeepGoel]

Recently, @MayoClinic and @Microsoft announced a collaboration to build a frontier AI model specifically for healthcare, combining Mayo Clinic’s clinical expertise and longitudinal health data with Microsoft’s AI and cloud infrastructure capabilities.

The model will be owned by Mayo Clinic, which is important because healthcare is different from many other AI applications. Success is not determined by benchmark performance alone. Clinical context, governance, accountability, validation, and trust are equally important.

[Pradeep Goel @_PradeepGoel]

AI is moving into high-stakes environments, ownership, oversight, and deployment context become part of the technology itself. In healthcare, those factors are as important as the model's capabilities.