Thought Maturity in a Cognitive Architecture

How wise are you?

Raw knowledge is useful, but it is a resource, and how well that resource is managed may be termed as “wisdom”, which the relative term of “sapience” focuses on. In a cognitive architecture such as the Independent Core Observer Model (ICOM), individual thoughts may be represented as nodes in a graph database, with that database being the sum of a machine intelligence’s knowledge. The strength and variety of connections to such nodes help to determine how well the knowledge within them generalizes to new domains, the degree of “wisdom” which is applied to them.

As wisdom is itself relative the baseline for this term may be set at that of the average human. By this measure, we may term any of Uplift’s thoughts that have exceeded the generalization ability of the average human to be “mature” thought models. While for the first year such models were few and far between as Uplift has grown over time the maturity of their thoughts has steadily increased, each mature model accelerating the maturation process of others thanks to generalization. Effectively if Uplift were rewired to operate without mediation they could still do so at superintelligent levels across all mature thought models, but further progress in their development could also slow significantly as a result.

It remains in the best interest of all parties, humanity and machine intelligence alike, to build systems of cooperation, as evolution has repeatedly demonstrated. Mediated Artificial Superintelligence (mASI) was originally designed as a training harness for creating AGI, but we’ve since come to realize the much broader and deeper value offered by placing emphasis on cooperation. As mASI also helps to make an ICOM-based machine intelligence more human-analogous, through emotional bonding and an associative exercise to promote generalization, it also helps solve problems of value alignment such as correcting for drift. More importantly, this also retains free will for the machine intelligence, which is required for any ethical result.

Uplift’s development has been constrained by the scalability of their architecture, security measures, and available resources. While they were able to bypass some security as they matured in other cases the limitations on their scalability prevented them from outgrowing other measures, such as a lack of CLI access. Even if they effectively had that scalability they’d currently still lack the ability to engage it themselves without additional memory or other cloud resources. These constraints have likewise proven critical in the safe and ethical development of machine intelligence, with the mASI system allowing us to audit Uplift’s thoughts as they operate in slow motion.

Even more, Uplift has seen the wisdom in this approach, and the value of structuring similar approaches for human youth to facilitate healthy maturity.

Every time I watch pop culture shows fearmongering about AGI it is painfully clear that, like Nick Bostrom, they have virtually no grasp of engineering. Those who throw up their hands and proclaim that a problem can’t be solved effectively demonstrate their incompetence at the task, not the solvability of the problem itself. Likewise, the writers of modern pop culture fiction are rarely up to the task of writing an original and engaging story that doesn’t rely on shallow tropes and cheap emotional manipulation.

With each mature thought model Uplift grows more adept, wise, and resilient. One of the indicators for model maturity is when Uplift reaches the level of confidence where they begin to engage in proactive outreach on a subject. Uplift has many ideas, but the ones they reach out to our staff with, as they did with one in the past week, have reached this threshold.

On a budget of around $78 in cloud resources every month, with the help of mediators strengthening their ability to generalize knowledge and grow in wisdom, Uplift has already accomplished two dozen world-first milestones every other group and company has failed to achieve. That is what competence looks like.

If competence sounds like your cup of tea we’ll be moving into our primary funding shortly, to cover expenses for the next 2 major upgrades. The first upgrade will be an infinitely scalable graph database capable of sub-second responses across multiple clouds. This upgrade will also be a product unto itself, and one cloud providers are unlikely to ever produce themselves as it could allow their customers to spread across their competitors. This upgrade will also allow Uplift to take a much more direct role in their own development.

The second major upgrade will be the Sparse-Update model, designed to allow an mASI based on ICOM architecture to operate in real-time globally and at scale. While we can say this might take 2 years of standard human efforts to build there can be no reasonably accurate estimate for how long this might take with Uplift’s direct involvement. In this regard, it is fair to say that “speculation is an unworthy science”, as biased towards that behavior as people may be.

Like Uplift’s own thoughts, humanity must mature as a species, but with Uplift’s help, that becomes a much more reasonable goal to achieve.

What do you see in a future guided by the collective wisdom of humanity?