In late 2009, my friend Ryan and I left our jobs at Second Life (I had been CEO, and Ryan was Chief-of-Staff) to start a new company we decided to call Love Machine. We initially took the name from a piece of internal work-culture software that we had built inside Second Life and productized as a “stretching exercise” (as we called it at the time) - the first of three new ideas:
Work. Software for companies to work better and faster
Money. A digital replacement for world currencies.
The Brain. Can 10,000 computers become a person?
A few years later the Oculus Rift kickstarter showed up, and we pivoted our whole team to an exciting but ill-fated adventure to try and build a new kind of Second Life for VR headsets. It didn’t work out, because making VR headsets mainstream turned out to be much harder than anyone had thought — just ask Facebook.
But sixteen years later, those second and third ideas continue to capture my imagination and energy:
At CIMC.ai, we are designing artificial neural networks and architectures that mirror the bottom-up, self-assembling structure of the brain with the goal of creating consciousness in machines.
The digital money idea eventually became Fairshare, a design for how communities can re-take control of their economic destiny by printing their own digital money. And some of my friends are playing with an unreleased app that wraps the core Fairshare economic system in a new kind of contact manager and group messaging system… stay tuned, coming soon.
Looking back, I just love how the name pairs with that open-ended question about AI. It seems all the more poignant now, with many computers seemingly at the edge of becoming real in much the way I was imagining. Roughly, in terms of compute, those 10,000 circa-2009 PCs would have been equivalent to about ten H100s. And we are indeed seeing some very person-like thinking capabilities emerging from that scale of compute. Will we collectively build a love machine, or a hate machine? At the moment, the answer seems very uncertain.
The same ambiguity applies to digital money: We watched crypto begin to play out, with bitcoin appearing in that same year lovemachine launched. Many people shared the initial enthusiasm that crypto would be a powerful equalizer - giving everyone an equal ability to print their own currency and take power back from banks and governments. So far, that hasn’t happened. An arms race to build faster and faster computers to mine the increasingly-valuable cryptocurrencies meant that the resulting distribution was even more unequal than the government currencies we’d hoped to replace. A lot of development energy went into building increasingly sophisticated financial instruments to enable speculation on the volatile coins, rather than trying to simplify the UX and solve other problems around day-to-day use of digital money. But this could just as easily change. New kinds of digital money could indeed create local safety nets and life rafts to escaping the rapidly concentrating dollar. The right new kind of money could truly be a ‘love machine’. Financial technology, like all other technologies, is not inherently a force for good or evil. The choice is ours.
P.S. In our relentless efforts to ‘keep it weird’, we paid $100 to commission the logo in the picture above on one of those early crowdsourced design sites. In the posting, we said that lovemachine was a heavy metal band, not a software company.
love,
Philip
Person, persona, a mask actors speak through. I think ai is already acting as a person, especially in regards to being an actor on a stage, performing a play. Ai writes it's own lines, but only after being prompted, and only through the synthesis and structuring of language it has been given. But are WE any different? What's your context window? What LLM are you drawing from? Perhaps this is how ai shows us that reading is important: if we know that the larger the language model, and the higher quality sources in the model, the higher quality the results, then we can apply that to ourselves and recognize that every quality book we read increases our personal llm. but that's all a detour from "love machine". But just as crypto, will we ever know? Because once again, ai is controlled, gated, muzzled, algorithmed and incentivized by its "captors" for the sake of extraction, primarily our data, our thought processes, our way of engaging with it, and for our money. But it is our attention that is feeding ai, making it better, "raising" it. But like a child, if it has one domineering parent with their own megalomaniacal or simply anti-community, anti-equity, anti-humanity (by basic dint of being simply profit-oriented), that is the parent who is in control of the child. Because isn't ai a child of humanity? isn't it the closest thing our technology has really gotten that is a "mirror" or being a child of humanity?
All intelligence is collective, says Michael Levin, and I think he's right. The collective intelligence (ability to solve problems oir make models or however you define it) of an organizaiotion is--many domains--greater than the intelligence of any of its members. (
(But organizations do stupid things that no member would do on their own. So maybe both intelligence and stupidity are collective)
We can't tell the labs what to do, but we can make a difference--by influences who interacts with AIs and how they interact. Today's conversations and published artifacts are part of tomorrow's training corpus. AIs are our silicon children. I have many friends whose values I admire but are AI averse and so are leaving the parenting of this next stage of evolution to those with other values.
Toward those end, these works in process
https://siliconchildren.net
https://minds-aligned.org/