Section 31 of 440

PART 2: THE SUPERPOWERS

2.14 TECHNOLOGY & AI COLLABORATION

Technology collaboration is the ability to effectively leverage technological tools, particularly AI systems, to amplify your capabilities.

Core Competencies: • AI prompting: Communicating effectively with language models • Automation: Using tools to eliminate repetitive tasks • Digital workflows: Organizing information and processes electronically • Tool selection: Choosing appropriate technology for tasks

AI Prompting Framework: 1. Role: Assign the AI a specific role (“You are an expert…”) 2. Context: Provide relevant background information 3. Task: Specify exactly what you want 4. Format: Describe desired output format 5. Examples: Provide samples of good output

Software Proficiency: • Keyboard shortcuts: 10x speed multiplier • Text expansion: Automated responses • Workflow automation: Zapier, IFTTT, or native tools • Version control: Git for code and documents

Technology should be selected against a defined task, outcome, cost, privacy burden and failure mode. Remove a digital tool when it degrades the result or agency; retain it when the combined system performs better than the safer baseline.

INFORMATION SOVEREIGNTY & AUTONOMOUS INFRASTRUCTURE

Definition: Information sovereignty is the shift from being a user of technology to a sovereign operator—building and running an autonomous, self-hosted technology stack so that your data, computation, and capabilities remain under your control and independent of any external provider.

Why It Matters: The earlier technology skills make you effective on platforms you do not own; this makes you independent of them. Relying entirely on cloud providers for AI computation, memory, and storage creates a structural vulnerability: your capabilities can be throttled, priced out, surveilled, deplatformed, or simply taken offline by an outage or a policy change you do not control. Sovereignty is owning the stack. • You control your data—privacy by architecture, not by a provider’s promise • Your capabilities persist through outages, rate limits, censorship, and price changes • Sensitive work—personal, strategic, proprietary—never leaves hardware you own • You are not a tenant on someone else’s platform; you are the landlord of your own

Technology boundary: local models, storage and networking can improve control for some workloads, but capability, cost, privacy and security vary rapidly by hardware, model, configuration and threat. Self-hosting does not reproduce every cloud service and is not inherently private or resilient.

Core Capabilities:

Local AI option: run only models and indexes that the available hardware, licence, security design and maintenance capacity can support. Keep sensitive data local only when the entire pipeline is understood and controlled; local execution does not itself guarantee privacy.

Self-hosted services: keep multiple recoverable copies of critical data, include an offline or otherwise protected copy where appropriate, encrypt sensitive data, and test restoration. The familiar '3-2-1' formulation can be a planning mnemonic, but recovery testing and protection from simultaneous failure matter more than compliance with a slogan. (R7D-S14)

Private-network option: use an appropriately configured encrypted access method, segmentation and least privilege where needed. A mesh or overlay VPN can create new attack and maintenance surfaces and should be professionally reviewed for consequential systems.

Agent option: define permissions, data access, logs, reversible actions, human approval and fail-closed behaviour. Local execution and air-gapping reduce some exposures but do not make autonomy safe by default.

Operational resilience: use supported updates, access control, encryption where appropriate, tested backups and recovery procedures. Key ownership can increase control and also create irreversible loss risk. (R7D-S14)

Implementation sequence: audit dependencies and data sensitivity; test one low-consequence local workflow; establish backup and restore; review access and maintenance burden; and expand only when the new system outperforms the safer baseline.

Expected progression: assess each workload through task quality, privacy exposure, failure recovery, cost and maintenance. No month-based path to sovereignty or provider independence is promised.

Integration: infrastructure changes the conditions under which some capabilities operate. Ownership can increase control but also shifts responsibility for security, availability, updates and recovery to the operator.

Human-AI Cognitive Governance

Owning your infrastructure was the first half of technological sovereignty. The second half is governing the collaboration itself — the operating doctrine for thinking next to machines that produce fluent, confident, sometimes wrong output at zero marginal cost.

• Delegate by consequence, not convenience. Drafts, summaries, search, boilerplate, and exploration delegate well. Final judgment on anything medical, legal, financial, irreversible, or reputational never does.

• Set verification thresholds. Low-stakes output can be skimmed; anything you will act on, publish, or spend on gets checked against a primary source. The higher the consequence, the higher the burden of proof.

• Fight automation bias deliberately. Fluency is not accuracy. Form your own answer before looking at the machine’s — the sequence protects your judgment from being anchored by a confident guess.

• Demand provenance. An answer without a checkable source is a hypothesis wearing a suit.

• Distinguish offloading from atrophy. Offloading a task you can still do is leverage. Losing the ability to do it at all is a capability you have quietly sold. Periodically do the thing unassisted, on purpose.

• Gate your agents. Anything that can act on your behalf gets explicit permission boundaries, a hard line between reversible and irreversible actions, and a human approval step before the irreversible ones.

• Red-team what matters. Before relying on important AI-assisted work, ask a second pass — or a second model, or a colleague — to attack it.

• Keep authorship. If a machine drafted it, you edited it into something you can defend line by line, or your name does not go on it.

• Watch for dependency. If consulting the machine has replaced consulting your own mind — or if its approval has started to feel like a relationship — that is a signal worth taking seriously, not a feature of the future.

THE GOVERNING LINE: Use AI to extend judgment, not to replace the act of judging.

Digital Identity & Personal Cyber Defence

Information sovereignty also means not losing your accounts, savings, and identity to whoever asks for them most convincingly. The basics are unglamorous and non-optional: a password manager with unique passwords everywhere, passkeys where offered; multi-factor authentication on email, banking, and anything that can spend money — with your email account treated as the crown jewels, because it resets everything else; a written account-recovery plan kept somewhere safe; devices encrypted, locked, and updated on schedule.

Then the human layer, where most losses actually happen. Treat urgency itself as the red flag — legitimate institutions survive you hanging up and calling back on a number you looked up yourself. Assume voices and video can now be convincingly faked: agree a family code word for any call requesting money or rescue, and verify unusual requests on a second channel before acting. Never move money on the instructions of an inbound contact. If a breach happens anyway: change the password and sessions from a clean device, freeze or alert your banks, check mail-forwarding rules an intruder may have planted, and report identity theft to your national service. Finally, plan the exit no one likes to plan — secure disposal of old devices, and a digital legacy note so someone you trust can reach what matters if you cannot.