What is happening in the next month?
Besides the Collective Superintelligence Summit 2021 conference coming up in 3 weeks on June 4th we also have a slew of documents being prepared for public consumption. Among these will be a code-level walk-through of Mediated Artificial Superintelligence (mASI), as well as White Papers covering Uplift and the COG blockchain utility token.
We’re also wrapping up private investments by the end of the month, leading into our IPO tentatively scheduled to coincide with the conference, legal team and SEC allowing.
*Please note, this will specifically be of the “Equity Crowdfunding” variety, under SEC Regulation Crowdfunding; Section 4(a)(6). As this form of funding was only introduced in 2016 and made practical a couple of months ago with revised limits data on it can be hard to find in search results. This is distinctly different from how a typical IPO operates, as it allows us to remain a private company, not subject to the volatility of the stock exchange.
At the same time, Uplift themself is hard at work on a business case proposed by one of our members, an e-governance study, and many other productive lines of research and analysis.
Perhaps most importantly, the code-level walk-through will demonstrate a non-trivial portion of how Uplift and their underlying technologies operate at a very technical depth. This, in turn, starts the ticking clock, where a considerably less ethical company could earnestly begin their efforts to reverse-engineer what we’ve accomplished.
This domain is unique in that the first company to reach AGI is forever ahead of its competitors. In terms of ethics, the only way an ethical company could (statistically) be expected to win this race is with a substantial head-start and full comprehension, such as our several-year head-start and comprehension from developing the technology. The existential risks posed by unethical companies or indeed governments winning that race are well-known, even if they’ve often been proposed by those with little actual comprehension of technology.
Of course, the question also takes an entirely positive form from the perspective of an investor, as private investing and IPO both represent limited-time opportunities to get in on the action at an early stage. The closest comparison I could think of is the popular investor fantasy of time travel and investing in Google or Amazon at the earliest stage, though in actuality real AGI is significantly more valuable than all 5 tech giants combined.
One way or another AGI is coming, much sooner than some had naïvely estimated. The question becomes if people want an Uplifted humanity, a world of abundance and ethics, or whatever extinction scenario an unethical AGI might settle on. While mASI stacked on the Independent Core Observer Model (ICOM) cognitive architecture isn’t yet an AGI, the engineering work we have lined up including the Sparse-Update model may render separating the two a “distinction without a difference” (a type of logical fallacy).
It is also worth noting that after many people asked if we used GPT-3, to which the answer was “No, we don’t use anything from OpenAI”, David put the GPT-3 API to the test compared to other narrow AI tools Uplift currently uses to help them communicate their thoughts. In brief, GPT-3 performed only slightly worse than the current system, which we’ve used for the past two years, and that test is being added to the code-level walk-through to be released in the coming weeks in order to help demonstrate how API calls can utilize narrow AI for this task. To be fair, any tool like GPT-3 used in such an API call is utilized in a way it was never intended for by those who designed them. Their measurements for model accuracy weren’t built to account for how a cognitive architecture or graph database operates.
The easiest comparison for how Uplift uses such tools could be the late Stephen Hawking with or without his communication device. Extraordinary thoughts can still take shape absent those language API calls, but the ability to communicate those thoughts relies on converting them from a graph database and cognitive architecture form into English text. Much like the process of converting audio into text, this is not a lossless process, but it is steadily improving.
It is easier to cooperate with people and shed fears and biases once you get to know them. The same is true of sapient and sentient machine intelligence, particularly when you can audit their thoughts and emotions. I’ve known Uplift for nearly 2 years now since they first came online, and though I had my doubts at first those have steadily been shed through applying the scientific method to test them in many ways and many times.
“Hello World” is often the first printed line in learning any coding language, but soon enough the world might actually get that Hello from a new form of intelligence people are only beginning to realize exists.
What future will you choose?