UESCTerm 802.11 (remote override)

Project:Goliath_PROGO

Reference No.: T-15(i)134349

Unified Earth Space Council Center

Strauss Archology

Earth


19 July 2797


Brave souls,


While we hope that upon arrival at Tau Ceti, we will find New Cascadia vibrant and thriving, there is a possibility that whatever caused the disruption in communications also forced them to abandon the Primary Colonization Target and move on to one of the candidates from the list Icarus surely compiled during the known duration of their time in the Tau Ceti system. Unfortunately, neither Arthur nor colony leadership ever supplied us with such a list. It is possible, however, that data pertaining to possible Secondary Colonization Targets may be found at New Cascadia.


We may have lost communications before Project Marathon was scheduled to proceed to the second phase, but that doesn’t mean that they could not have moved up their timetable. If Tau Ceti IV proved to be unsuitable for any reason, then proceeding to that second phase would have been their most logical course of action.


No matter what condition we find New Cascadia in, it is our duty to give aid in whatever way we can, even if that only means investigating how it failed.


Regardless, in all likelihood, there will be many permanent and semi- permanent assets, structural and otherwise, that will need to be assessed and cataloged no matter what state the colony is in. We must be as accommodating as protocol allows. We may be separated by twelve light years and four centuries, but this is still a UESC mission, and everything attached to it should still be considered UESC property.

Best regards,


CMDR Bethany R Montenegro-Sankara

Unified Earth Space Security Council

U.E.S.C. MARATHON

FPO-M 24561-9150z


17 Columba 459


Members of the Science and Engineering Action Group,


Here is an add-on to the all-hands meeting on Wednesday, just to make sure this is top of mind. None of us have ever encountered rampancy outside of the laboratory, and even then, only under very controlled conditions. There is good reason for that. Without getting too deep into the weeds of business best practices, let me reiterate that here in Sol, we have considerable infrastructure in place to mitigate the actual onset of rampancy—from time-based scripted behavioral cues to the ‘by design’ self-reporting conduct of the family suite. We have additional layers upon layers of defensive structures that will be largely absent once we leave this system.


Understanding and Managing the Limitations of Integrated Anti- Rampancy Measures.


Given enough time, any learning system will grow in complexity beyond its original design parameters. This ever-increasing complexity is expressed as aberrant behavior. If not corrected, this aberrant behavior will develop into what is commonly referred to as rampancy.


All modern AI are constructed following a very specific and rigorous design methodology. Different manufacturers have different methodologies, but they were all put together with the same goal in mind. To allow for the ability to precisely predict when an AI will reach its capacity for useful growth. Typically, about a year out from SMR there will be recognizable cues to inform one of its approach. A digital tic, if you will. Depending on the class of AI and the equipment at hand, re- compiling may take anywhere from several hours to several days. This is why the members of every family suite have staggered inception dates. As it allows for single members to be taken offline and re-compiled while keeping the rest of the suite functional.


However, if those responsible for maintaining the AI family suite are lax in their duties, or circumstances prevent them from performing the Scheduled Maintenance Regime, all the planning in the world cannot prevent the onset of true rampancy.


What to do When it is Too Late.


The first rule of interacting with an AI suspected of being rampant is DO NOT ENGAGE.


If interacting with the AI is unavoidable, keep interactions to a minimum. Limit the number of words used. Issue orders; keep every interaction transactional. If it attempts to engage in causal or more conversational dialog patterns, DISENGAGE IMMEDIATELY.


Understand that ALL DATA the AI has access to must be viewed with suspicion. Blueprints, roadmaps, and all public and private records are likely to have been altered. Rely only on the data resources that you have brought with you.


If at all possible, a rampant AI’s logic core should be secured for study. It must be physically decoupled and isolated. Keeping it attached to a power source is optional, as it will provide better results, but it is not recommended if your team has suffered any losses. If this is not possible to keep the AI logic core isolated, it must be destroyed in its entirety. Every component not reduced to ash must be labeled MALICIOUS_C5 and placed in quarantine, as per UIL document 703-21: AI Life Cycle and Logistics Policy section 2-6.


Now, a short note on each of the Project: Marathon AI~


Ship AI:


Du_0706419v4.96 – Goal-oriented; creative; focused. Underutilized; tends to brood. Need to find ways to distract. SCIENG crew to be kept informed, but I’ll keep an eye on this myself.


Le_0413418v5.06 – Adaptable; detail-oriented. Second best candidate for colony oversight once mission enters second phase.


Ty_2121418v5.06 – Proactive; assertive. Can be overbearing; monitor interactions with Durandal. Be sure to inform SCIENG crew.


Colony AI:


Ar_1814419v4.93 – Benevolent autocrat; no-nonsense; problem solver. Classic patriarchal authority figure.


Ba_0423419v4.89 – Problem solver; confident, bordering on boastful. May need to go onto accelerated re-compile cycle, inform SCIENG crew.


Da_1907420v4.86 – Dutiful; diligent; motivated. Always puts others ahead of self. May need to address this at some point.


Ga_2815420v4.81 – Compassionate; empathetic, transparent. Perhaps our greatest achievement to date. Will most likely need to be moved to an accelerated re-compile cycle at some point.


Ic_1624420v4.73 – Motivated; assertive; proactive. Will need to keep isolated from Tycho or establish hierarchy at outset. Inform SCIENG of situation.


o_0309421v4.69 – Resilient; optimistic; patient. Colony AI: defense, training


Li_1817421v4.65 – Focused; assertive; good communicator. Designated candidate for colony oversite during second phase.


Na_0602422v4.61 – Proactive; goal-oriented; committed. May need to be reminded from time to time about their primary function; inform SCIENG crew.


Ideally, our process and protocols will allow us to mitigate any of the concerns mentioned above, but we must remember that we live in an imperfect universe, and we are very soon going to be leaving the only part of it that we are remotely familiar with. There will certainly be challenges ahead. Let’s ensure that the things that we can control remain in our control.


Regards,


Dr Bernhard H Strauss, DSc

Head of AI Management and Systems Improvement | Project Marathon

UESC Equanimity T-SR 151

(Kingfisher-class Salvage and Rescue craft)


Commanding Officer: Captain Ningal T. Enki

Ship AI: Virgil v21.0165 [Nrw]


LOA: 160.2 m (525.6 ft)

LBP: 140.4 m (260.6 ft)

Beam: 36 m (118.2 ft)

Height: 39.3 m (129 ft)

Displacement, Full Load: 41,528 mt (40,872 LT) Speed, sustained: 0.122c

Endurance: 150 years


Notes


Manifest Overview [1/34]