Agency Over AI Agents - World Design Policy Conference 2024

We've just entered the AI Agentic Age. "Agency Over AI Agents" - is a critical exploration of AI governance at a pivotal moment in technological history. As we stand at the crossroads of unprecedented AI advancement and massive AI deregulation. We'll tackle the urgent question; How do we maintain human agency while harnessing AI's potential?

As part of the World Design Policy Conference for the World Design Capital of San Diego + Tijuana, we'll examine real world strategies for balancing innovation with responsibility. You'll discover why focusing on risk management over mere compliance is becoming essential, and how communities are adapting to the AI-driven future. We'll explore practical frameworks for AI governance that put humans in the driver's seat, not just as passive observers.

With the looming deregulation and AI capabilities expanding daily, understanding these dynamics isn't just academic - it's crucial for anyone involved in technology, policy, or business. Join us as we map out a path forward that preserves human agency while embracing AI's transformative power.

Let's ensure we're steering AI's development, rather than letting it steer us.

Transcript

1. AGENCY OVER AI AGENTS @kyleledbetter - AskDream.ai
2. THE AI AGENTIC AGE A new era of autonomous systems can independently perceive, reason, act, and learn to accomplish complex multi-step tasks with minimal human intervention. WRITER CODER LEGAL TASKS CREATIVE a11y BUGS
3. MASSIVE DEREGULATION The outcome of our decision as a nation. "We will repeal Joe Biden’s dangerous Executive Order that hinders AI Innovation, and imposes Radical Leftwing ideas on the development of this technology"
4. INNOVATION CONSEQUENCES "Even with the best of intentions, without external oversight, this kind of self-regulation will end up unenforceable" "...The relentless innovation of US companies disrupts and reshapes global industries. Europe must adapt to the opportunities and challenges of AI, or risk ruin." Alex Karp, CEO of Palantir Technologies Helen Toner, Former OpenAI Board Member
5. INNOVATION & GROWTH Unregulated potential for explosive advancement.
6. ABUSE & MISUSE Unregulated potential abusing this incredible power. 𝕏 Government Surveillance Risks • AI-powered mass surveillance through integrated public/private cameras • Predictive policing targeting specific communities • Health data monitoring and reproductive healthcare tracking • Social media analysis for "pre-crime" identification Corporate Exploitation • Unregulated AI hiring systems with hidden biases • Predatory lending algorithms targeting vulnerable populations • Insurance discrimination based on AI profiling • Medical care rationing through automated systems Immigration Targeting • AI-powered continuous surveillance at borders • Automated deportation decisions without human review • Biometric tracking systems linking multiple databases • Social media monitoring for family/connection mapping
7. HUMANS IN THE LOOP Humans need to be involved depending on risk level of use case. 01 HITL EVERY DECISION 04 HOOTL FULLY AUTONOMOUS 02 HOVL or HATL STRATEGIC DECISIONS 03 HOTL MOST DECISIONS 01 Human IN The Loop HITL Humans are directly involved in AI decisions, reviewing and approving each step. ex: An AI-assisted medical diagnosis where a doctor reviews and confirms each diagnosis suggestion before action is taken. 02 Human OVER/ABOVE The Loop HOVL/HATL Humans set high-level goals and guidelines for AI, but the system operates independently within those parameters. ex: Financial trading algorithms where humans define risk limits and strategies, but the AI makes trade decisions independently. 03 Human ON The Loop HOTL Humans oversee AI operations with the ability to intervene if necessary, but do not review every decision. ex: Autonomous driving systems monitored by a remote operator who can take control in case of unusual or risky situations. 04 Human OUT Of The Loop HOOTL AI operates completely independently without human oversight. ex: Predictive maintenance systems in manufacturing that autonomously schedule repairs without human intervention.
8. POLICIES WITHIN OUR CONTROL Internal and personal policies are needed in lieu of regulations. Corporate Internal Policies Data Fortress Protocol Mandatory data encryption for all AI interactions Strict prohibition on inputting sensitive data into public AI tools Required use of enterprise-grade AI tools with proper security certifications AI Usage Boundaries Clear list of approved/prohibited AI tools Specific guidelines on what types of work can/cannot use AI Required human review of all AI-generated content before external use Accountability Framework Designated AI oversight committee Mandatory documentation of all AI use cases Regular audits of AI outputs and impacts Personal Protection Policies Digital Identity Shield Minimal sharing of personal images/voice recordings online Use of watermarks on personal content Regular monitoring for unauthorized AI use of personal data Data Minimization Strategy Opt-out of AI training data collection whenever possible Use of privacy-focused AI tools only Regular deletion of personal data from AI services
9. RISK OVER (FEDERAL) COMPLIANCE AI Governance focus will shift to risk identification & mitigation
10. STATE REGULATIONS As the country deregulates, states will lead the way. SB24-205 Consumer Protections for Artificial Intelligence SB-149 Artificial Intelligence Amendments SB-942 California AI Transparency Act NYC-LL144 Automated Employment Decision Tools
11. COMMUNITY & IRL In times of AI deregulation and rapid technological change, community-based approaches become crucial for protecting and empowering citizens. Strong community networks and collaborative structures offer essential safeguards and opportunities that government oversight may no longer provide.
12. THE ROLE OF POLICY MAKERS As artificial intelligence rapidly transforms society, policymakers face unprecedented challenges in balancing innovation with public protection. Their role becomes even more crucial during periods of reduced oversight and regulatory change. COLLABORATE WITH EXPERTS DEVELOP NEW STANDARDS IDENTIFY NEW RISKS RESEARCH NEW USE CASES
13. AGENCY OVER AI We need to ensure people we trust are in the driver's seat. askdream.ai If we rely too heavily on AI to do our thinking for us, we may lose some of our ability to think critically and creatively. We may also become less able to learn new things and adapt to change. The ability to choose gives us agency over AI, allowing us to wield it to our benefit. The key is to find a balance between using AI to automate tasks and using our own brains to think and learn.
14. Thank you! askdream.ai @kyleledbetter askdream.ai