World Solve

Public agencies will buy models, datasets, inference services, and automated workflows from vendors whose systems may not be easy to inspect. The unresolved problem is creating contract standards that make public AI procurement auditable, portable, and lawfully contestable. The challenge is especially acute in informal settlements, where legacy systems, coordination failures, and weak oversight can magnify harm.

States may invest in national AI systems for education, law, defense, planning, and public administration. The unresolved problem is ensuring such systems remain bounded by law, reviewable by citizens, and resistant to capture by narrow interests. The challenge is especially acute in informal settlements, where legacy systems, coordination failures, and weak oversight can magnify harm.

Frontier AI may become essential infrastructure controlled by a small number of governments or companies. The unresolved task is developing ownership, access, competition, and public-interest models that prevent private or geopolitical concentration from determining global outcomes. The design must remain functional when standards, trade routes, data access, and diplomatic cooperation are unreliable. The research opportunity is to define measurable success criteria, test the approach in realistic settings, identify failure modes, and create a pathway from prototype to accountable deployment.

Future governments may depend on private firms for cloud services, AI tools, digital identity, and critical infrastructure software. The unresolved problem is designing procurement rules that preserve interoperability, portability, and long-term public control. The challenge is especially acute in informal settlements, where legacy systems, coordination failures, and weak oversight can magnify harm.

Future infrastructure will increasingly rely on systems capable of making decisions without immediate human supervision. The unresolved problem is creating architectures that can prove what they did, explain uncertainty, fail safely, and remain controllable during abnormal conditions. The solution must produce useful coordination and accountability while minimizing collection of sensitive personal and behavioral data. The research opportunity is to define measurable success criteria, test the approach in realistic settings, identify failure modes, and create a pathway from prototype to accountable deployment.

Future infrastructure will increasingly rely on systems capable of making decisions without immediate human supervision. The unresolved problem is creating architectures that can prove what they did, explain uncertainty, fail safely, and remain controllable during abnormal conditions. The design must remain functional when standards, trade routes, data access, and diplomatic cooperation are unreliable. The research opportunity is to define measurable success criteria, test the approach in realistic settings, identify failure modes, and create a pathway from prototype to accountable deployment.

Future infrastructure will increasingly rely on systems capable of making decisions without immediate human supervision. The unresolved problem is creating architectures that can prove what they did, explain uncertainty, fail safely, and remain controllable during abnormal conditions. The solution must work with limited capital, intermittent services, and local maintenance capacity. The research opportunity is to define measurable success criteria, test the approach in realistic settings, identify failure modes, and create a pathway from prototype to accountable deployment.

Governments will increasingly rely on automated systems to allocate benefits, flag risks, prioritize inspections, and recommend enforcement. The unresolved problem is creating governance structures that make these systems transparent, contestable, and accountable without disabling public service delivery. The challenge is especially acute in informal settlements, where legacy systems, coordination failures, and weak oversight can magnify harm.

Automated trading algorithms can sometimes trigger rapid, extreme price swings in smaller or less liquid markets, especially during unusual conditions, without any single actor intending market harm. Building safeguards against this kind of systemic risk remains incomplete.

Businesses seeking to cut corners could potentially use AI to generate fabricated but convincing compliance records for regulatory audits, and auditors don't always have the tools to catch increasingly sophisticated fake documentation. Strengthening audit verification remains an unresolved challenge.

Networks of AI-assisted fake dating profiles can be deployed at scale to build trust with vulnerable users before extracting money or personal information. Detecting these coordinated networks before real harm occurs remains an unresolved platform challenge.

Fabricated reviewer identities, complete with fake credentials, have occasionally been used to manipulate the peer review process in favor of certain papers. Verifying reviewer authenticity more rigorously remains an unresolved challenge for academic publishing.

Safety filters designed to block clearly harmful requests can sometimes be bypassed through indirect or creatively phrased questions that still elicit dangerous information. Closing this gap while preserving the system's general usefulness remains an unresolved balancing act.

Digital contracts can potentially be modified after signing in ways that aren't always immediately detectable without specialized forensic review. Building this verification directly into standard contract systems remains incompletely adopted.

Some counterfeit or unsafe products carry fabricated safety certification marks that can be difficult for both consumers and regulators to quickly verify as fraudulent. Strengthening certification verification systems remains an unresolved consumer protection issue.

Attackers can automatically test massive numbers of previously leaked username and password combinations across many websites to find accounts where the same credentials still work. Reducing the effectiveness of this widespread technique remains an unresolved security challenge.

Some financial institutions still use voice recognition as an authentication method, but increasingly realistic AI voice cloning threatens to undermine this security layer. Strengthening these systems against synthetic voice attacks remains an active, unresolved challenge.

AI chat systems can sometimes misread customer frustration or urgency, responding in ways that unintentionally worsen a situation rather than resolve it. Better emotional context understanding in these systems remains an unresolved design challenge.

Companies can use AI to generate legal language that's technically accurate but deliberately obscures consumers' actual rights, making it harder to understand what protections truly apply. Improving genuine clarity in these documents remains unresolved.

Automated pricing systems can sometimes raise prices sharply during shortages or emergencies based purely on demand signals, without regard for fairness or necessity. Building ethical guardrails directly into these systems remains inconsistently implemented.