Diplostack #7 - Cogito, Ergo Sum

A new sentient intelligence is only a handful of breakthroughs away, of our own making; how do we retain its humanism and our humanity while reaping the rewards of this advance? What should NZ do?

I think, therefore I am -Descartes, 1637

These are but a few of the many questions posed at Diplosphere’s talk at the NZ Parliament Artificial General Intelligence: Have we Opened Pandora's Box? On stage with Asa Cox (Arcanum AI CEO/founder), Nina Obermaier (EU Ambassador to NZ), Scott Houston (ex-Green Button, author, entrepreneur, investor), Peter Griffen (Businessdesk) with Maty Nikkhou-O’Brien, founder of Diplosphere, presiding. Comment below if you’d like to continue the debate.

One panelist asked the question: What is the tipping point for artificial intelligence becoming sentient (“I think, therefore I am”)? We do not have a general definition of Artificial General Intelligence yet. But it matters. We will meet for the first time an alien intelligence of our own making, perhaps vastly superior to us in some ways.

You can find the podcast here:

It is not hyperbole to give a 50-50 chance of an artificial general intelligence in our midst within the next ten years, by the early 2030s (link). This intelligence will be unlike our intelligence but an alien intelligence that will be able to reason and converse with us humans, trained as has been within a “box” - massive clusters of computers using a gigantic corpus of data available from the internet. This intelligence we will find hard to define, or even call it intelligent. The way it works, under the hood, is still very much a blackbox. But it is coming nevertheless. And it will change everything. The genie is out of the bottle. Pandora has opened the box. Or as one speaker put it, Prometheus has stolen fire from the gods. It is up to us humans alive now at this moment to get the controls right before we lose control, if we can. We owe it to future generations. Many in the room agreed that this development - artificial general intelligence, quickly followed by “super-intelligence” - is a global challenge on par with or greater than climate change.

AI optimists were the majority with 3 of 5 panelists, but caution was certainly expressed throughout, and legislation welcomed. The EU Ambassador’s contribution on the EU’s proposed AI Act (link), likely to come into force in 2024, was welcomed as a guide-star for our own policymakers. But many lamented the slow action thus far (Our politicians need to wake up to the power of AI Peter Griffen, BusinessDesk). The line between fake and real images, audio, text will be blurred. The impact on jobs in the short-term will be tremendous, with one audience member questioning where the entry level university graduate jobs will be? Law, accounting, tech all will be affected. And that is just with the current state. Who knows what the next five years will bring?

We are close to the tipping point, a point of no return, it is maybe a handful of breakthroughs away. It is incumbent on us to move quickly now to set things up for success rather than a path of failure. One thing is that we don’t understand what is happening within these LLMs, they are “black boxes” of inscrutable matrices. One cannot determine in advance the output with 100% accuracy as one can with traditional programming and algorithms. This is quite different.

“Regulate us”, is the call from tech leaders like OpenAI’s Sam Altman and Google’s Sundar Pichai, but they don’t want it to stifle innovation, as says the EU. But the reality on the ground is that 10s of billions of dollars is already being mobilised to better, faster, cheaper AI.

For the moment, the metaphor that may be more apt is indeed “Prometheus stealing fire from the gods”. Let’s not transmute it into Opening Pandora’s Box.

Some references / pertinent stats (search Google, references to come)

• AI could see a 7% lift in productivity

• Goldman Sachs report on jobs - 25% across Europe/US lost?

• Cross party AI committee has been setup in NZ by Judith Collins

• Sam Altman testimony to congress

• Max Tegmark - Life 3.0

• Carl Bolstom - Superintelligence book, Singularity book

• G7 looking at this, cross party committee led by Judith Colilns (NZ)

• NZX using ChatGPT for market reports already

• EU AI Act - Bell Gully

• Worst case scenario 10% of AI scientists think disaster could result

• There is a call for an AI Pause (Elon Musk) at least of training

• Podcasts

  • Peter Griffen BusinessDesk

  • Lix Fridman - Sam Altman interview

  • Hard Fork podcast - game characters gossiping with each other

YouTube recording of event: