The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Oracle chairman and chief technology officer Larry Ellison made the claim at Oracle AI World in Las Vegas on October 14, 2025. His argument was not that artificial intelligence has already improved society, but that AI connected to private enterprise data could improve healthcare, food production, public safety and business operations.
That vision is also a commercial pitch: Oracle wants to provide the databases, cloud infrastructure, AI models, applications and automated workflows needed to put enterprise AI into production. The potential is real, but so are the unresolved questions around privacy, reliability, employment, cost and accountability.
What Ellison said at Oracle AI World
Oracle’s first customer event under the AI World name marked a shift from its former CloudWorld branding. In his keynote, Ellison presented AI as a force that could make the world better by helping organizations understand information, automate decisions and improve outcomes.
Free tools Windows power users keep installed
One-click scans. No signup required.
Coverage of the keynote described his examples as including medical diagnosis, patient monitoring, agriculture, robotics, fraud prevention, public safety and administrative automation. Oracle’s own AI Data Platform announcement framed the technology around combining cloud infrastructure, databases, generative AI and enterprise workflows.
#1 Best Overall
The important distinction is between a forecast and an established result. Ellison predicted that AI will create a better world. The keynote demonstrations showed possible applications. Neither, by itself, proves that Oracle’s systems have delivered better outcomes at population scale.
Oracle’s central theory: useful AI needs private data
Oracle’s pitch rests on a distinction between training a general-purpose model on public information and applying a model to current, private, institution-specific information.
A hospital needs to work with clinical records, imaging, staffing data and insurance processes. A manufacturer needs inventory, supplier and maintenance information. A bank needs transaction records and fraud signals. A model trained on public internet data cannot automatically know the latest internal facts or policies.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Oracle’s proposed architecture generally works like this:
- Enterprise records and documents are stored or connected.
- The information is semantically enriched and converted into vectors for similarity search.
- A model retrieves relevant private information.
- The model generates an answer, recommendation or summary.
- An AI agent may take the next step in a multi-stage workflow.
- Human approvals, permissions and audit controls determine whether the action is accepted.
This is closely related to retrieval-augmented generation and agent automation. Connecting a model to private data can make answers more relevant, but it does not guarantee that the underlying records are accurate, complete or lawfully accessed. It also does not eliminate hallucinations or bad decisions.
What “a better world” meant in practice
Healthcare
Oracle’s keynote examples included AI-assisted medical imaging, diagnosis, patient monitoring, connected ambulances, information sharing between healthcare systems and automation involving providers, insurers and payers. Other examples included predicting reimbursement and hospital financing, detecting disease through sensors and genomic analysis, and using machine vision and precise robotic movement in surgery.
These applications could address genuine problems. Faster access to relevant records might reduce administrative delays. Better monitoring could help clinicians identify deterioration earlier. Automating repetitive insurance and billing work could reduce waste.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
But the wording matters. A keynote demonstration is not clinical evidence. Medical AI requires validation for the specific population, device, workflow and decision being supported. It also needs regulatory compliance, clinical oversight and clear responsibility when an output is wrong. “AI can assist diagnosis” is a much narrower and more defensible claim than “AI will improve healthcare.”
Agriculture and food production
Ellison highlighted robotic greenhouses, indoor and urban agriculture, crop engineering, reduced water use, nitrogen-fixing crops, improved yields, carbon-dioxide management and autonomous drones for agricultural and environmental monitoring. Oracle has also described healthcare and food production as part of its broader generative-AI strategy.
AI could help farmers monitor crops, optimize irrigation or identify disease earlier. Robotics may make controlled-environment agriculture more precise. However, higher yields and lower environmental impact are not automatic. Results depend on electricity use, capital costs, crop suitability, local infrastructure, labor and regulation. An energy-intensive greenhouse may solve one environmental problem while creating another.
Enterprise and administrative automation
Oracle’s most immediate commercial opportunity is business automation. AI agents can retrieve information from company records, summarize documents, generate or modify software, coordinate departments and automate multi-step processes involving customers, suppliers, insurers and regulators.
Recommended Free Tools
This may reduce repetitive work and allow employees to focus on higher-value tasks. It may also change which jobs and skills are needed. Productivity gains should therefore be measured for a particular task and workforce, rather than assumed from the presence of an AI feature.
Public safety and surveillance
Ellison has separately discussed AI analyzing footage from street cameras, police body cameras, vehicle cameras and doorbell cameras. He suggested that continuous recording and reporting could discourage misconduct or crime. Reports from TechCrunch and Ars Technica have explored this vision.
Monitoring can improve accountability in limited settings, but more cameras do not automatically produce more safety or justice. Continuous surveillance creates risks of false identification, function creep, biased enforcement, abuse by authorities, chilling effects and data breaches. This example exposes the gap between Ellison’s universal phrase “a better world” and the reality that the same technology may benefit one group while imposing costs on another.
What Oracle is actually selling
Behind the social vision is a broad enterprise technology stack:
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minute- Oracle Cloud Infrastructure, or OCI, for computing, storage and networking.
- OCI Generative AI for managed model access and AI application development.
- Autonomous AI Database for managed database workloads, vector search and private-data access.
- The AI Data Platform for ingestion, semantic enrichment, vector indexing and agentic applications.
- Oracle Cloud Applications with embedded AI features.
- Healthcare, finance and other enterprise software connected to automated workflows.
Oracle’s generative-AI offering emphasizes managed services, enterprise security and governance, application integration and access to multiple models. Its documentation lists integrations involving providers and model families including Cohere, Google, OpenAI, Anthropic, Hugging Face and Amazon, while warning that availability varies by product and deployment.
Oracle’s documented activation model says covered AI capabilities require customer opt-in rather than being enabled automatically. That is useful, but opt-in is not a complete governance system. Customers still need to control access, retention, model use, audit logs, human review and data residency.
The evidence behind the optimism
The evidence should be separated into distinct categories:
- Product facts: Oracle has announced and documented databases, cloud services, model integrations and AI development capabilities.
- Demonstrations: Keynotes can show what a system is designed to do, but not how reliably it performs across real-world conditions.
- Customer outcomes: These require independently documented results, including the baseline, costs, error rates and affected population.
- Medical or agricultural effectiveness: These require the relevant clinical, field or regulatory evidence.
- Ellison’s prediction: The claim that AI will create a better world remains a long-term forecast and value judgment.
In short, Oracle has a plausible architecture for enterprise AI. It has not established that buying that architecture guarantees better social outcomes.
The risks Oracle’s vision leaves unresolved
Bad data and overconfident answers
Private data can be outdated, duplicated, incomplete or biased. A retrieval system may select the wrong record or fail to find a critical one. A model may then produce a fluent answer that appears authoritative despite weak evidence.
Medical and financial consequences
An incorrect chatbot summary is not equivalent to an incorrect diagnosis, insurance decision, payment or treatment recommendation. As AI moves from drafting information to taking consequential actions, organizations need testing, approval thresholds, audit trails, rollback procedures and a named person or institution responsible for the result.
Privacy and security
Enterprise systems may contain health, financial, employment, biometric and personally identifiable information. Buyers should ask where data is processed, who can access it, whether it is used for model training, how long prompts and outputs are retained, whether model providers can change, how deployments are audited and what happens during an outage.
Employment and deskilling
AI may reduce repetitive work, but it can also reduce demand for specific tasks and roles. Organizations may transfer expertise from employees to vendors and models, leaving workers with less control over decisions they are still expected to explain.
Vendor concentration and lock-in
Oracle’s integrated approach can simplify procurement and deployment for companies already using Oracle databases, OCI or Oracle applications. It can also concentrate infrastructure, data, model access and governance with one vendor. Migration costs, proprietary interfaces and consumption-based billing may make switching difficult.
Best Value
Energy and infrastructure
AI requires data centers, electricity, cooling, networking and specialized hardware. Claimed benefits in healthcare or agriculture should be assessed against the full resource cost of operating the systems, not only the efficiency of the final workflow.
What this means for organizations and investors
Oracle’s approach may fit an organization that already has a substantial Oracle footprint, needs AI over private enterprise data and prefers a managed, integrated platform. It may be less suitable for a small team that needs a simple chatbot, a buyer seeking maximum portability, or an organization that wants open-source, self-hosted or cloud-neutral infrastructure.
Alternatives include Microsoft Azure, Google Cloud, Amazon Web Services, direct model APIs, open-source models and specialist healthcare, agricultural or fraud-detection vendors. The meaningful comparison is not which company makes the biggest promise. It is whether a solution offers the required data residency, security, model choice, retrieval quality, latency, auditability, human review, portability, regulatory support and total cost of ownership.
Oracle documents both on-demand inference and dedicated AI clusters. Prices vary by model, region, SKU, currency, commitment and contract; Oracle’s own pages display differing example rates. Its free-tier and trial offers, including advertised US$300 credits for 30 days, are useful for testing but do not establish the cost of sustained production workloads.
The better test for Ellison’s prediction
The useful question is not whether AI can produce beneficial outcomes. It clearly can in some tasks. The harder questions are:
- Are the benefits measured against a credible baseline?
- Are they distributed fairly among patients, workers, customers, governments and vendors?
- Can a failed decision be detected and reversed?
- Can people understand and challenge consequential outputs?
- Who pays for the infrastructure, and who captures the productivity gain?
- Who is accountable when the system is wrong?
Ellison’s statement is best understood as a testable proposition: AI may improve outcomes when it is connected to high-quality data, constrained by governance and embedded in accountable workflows. The Oracle platform can supply many of those technical components, but it cannot guarantee the quality of the data, the fairness of the decisions or the wisdom of the institutions using it.
That makes “AI will create a better world” a compelling vision—and an unproven forecast, not a fact established by one keynote or one vendor’s product announcement.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
