On Track for ‘Breakthrough Discuss’

by Paul Gilster on April 14, 2016

Expect no new posts until Monday and a slowdown in comment moderation as I attend the Breakthrough Initiatives event in Palo Alto. As has been my practice at previous such events, I won’t try to cover the talks and discussions live. While ‘live blogging’ has its place and seems to work for things like product introductions, I don’t find it useful for scientific conferences, where I need to take lots of notes and think about the implications of what I’ve written. But I’ll come back from Stanford with plenty of things to talk about, and we’ll use all of next week sorting them out. Dinner tonight with Jim and Greg Benford, my son Miles, and my old friend Claudio Maccone, with whom I’ll be talking SETI, the FOCAL mission (to the Sun’s gravity lens), and the best methods of data retrieval for a mission like Breakthrough Starshot. Many more good discussions ahead. SETI is a big part of this conference, but Alpha Centauri looms large.


Andrew Palfreyman April 14, 2016 at 22:02

I am having trouble finding out how to attend. I’m a Bay Area lad.

Paul Gilster April 14, 2016 at 23:56

I’m afraid this one is an invitation-only event. Wish I could help!

Giulio Prisco April 15, 2016 at 2:11

We look forward to your report! Too bad they won’t webcast the talks. Or perhaps there is a webcast?

I am thinking about two ideas for the project and would like to know if they are discussed at the event. I guess the first will be discussed, but the second?

First, as the announcement notes, interstellar travel technology could permit achieving important intermediate milestones as it develops, and exploring the solar system and near interstellar space before Alpha Centauri. I wish to add my voice to those that suggest a mission to the (yet unconfirmed) Planet Nine as an intermediate Starshot milestone. Planet Nine is expected to be 0.01 light years (3 light days) away, which is far enough to be considered as an interstellar target but close enough to be reachable by precursor Starshot missions powered by preliminary, scaled-down versions of the laser propulsion system.

A Starshot fly-by mission to Planet Nine would permit acquiring useful data for future robotic (or even manned) missions, which could establish a first outpost in interstellar space. Establishing a base on or orbiting Planet Nine (if it’s really there) should be considered as an important priority for the first phase of interstellar exploration, especially if the base can double as a gravitational lens observatory.

Second, the StarChip should have the smartest possible on-board data processing software compatible with the hardware constraints. If smart software doesn’t fit on one StarChip, perhaps it could be implemented in a swarm of linked and co-operating StarChips.

It appears that really smart software is on the horizon: real thinking and feeling Artificial Intelligence (AI) of human (or more than human) level could be developed in a few decades. Sending a real AI instead of dumb unthinking software to Alpha Centauri would be equivalent to sending a person (think for example of Samantha, the AI in Her).

Even more interestingly, mind uploading technology, which could developed in this century, would permit sending human astronauts as software entities living in the StarChip processors. I have argued that an e-crew – a crew of human uploads implemented in solid-state electronic circuitry – doesn’t require air, water, food, medical care, or radiation shielding, can withstand extreme acceleration, and could be implemented in extremely miniaturized “manned” starships.

This seems a weird idea, but – as hinted at by theoretical physicist Avi Loeb in the announcement’s Q/A session shown in the video – post-biological life could be common among the stars. Perhaps, as Martin Rees and others think, advanced civilizations are post-biological. Hawking himself argued that intelligent machines based on mechanical and electronic components, rather than macromolecules, could eventually replace DNA based life, just as DNA may have replaced an earlier form of life.

Perhaps we could send AIs and uploads to the stars, combining interstellar expansion with post-biological evolution.

Robert April 15, 2016 at 12:48

We debated the possibilities of real AI a few weeks ago so I won’t rehash that quite yet but your post makes me think of another possibility. An advanced self replicating technology, possibly nanotechnology based, could be sent to explore and develop uninhibited star systems. By develop I mean not merely an infrastructure to support a human civilization, but the ability to create from raw information and chemicals, human beings. Once a few individuals are successfully grown, nature can take its course such that it takes a relatively short few centuries to go from a few to millions. Such a system can be propagated throuout the galaxy at a significant fraction of light speed. Then information can be sent in vast highways as to the nature of the galaxy and the progress of these civilizations as numerous as the stars.

George King April 15, 2016 at 22:58

Can hear them now . . . “what happens in an uninhibited star system, stays in an uninhibited star system.”

If the darn spell checkers would just stop auto-rewriting what we write, lol.

Not sure if it would be easier or harder to send frozen embryos — as they did in the Interstellar film plot line — in comparison to brewing up humans from scratch.

Either way, it would bypass the huge hurdles presented to sending grown human explorers and settlers across interstellar distances.

George King April 15, 2016 at 23:26

If they d0 grow them directly from scratch on site, humans formed from amino acids found in asteroids inhabiting newly terraformed worlds, Joni Mitchell’s line from “Woodstock” will be all the more apropos:

We are stardust
Billion year old carbon

Not necessarily back to the garden, but a new dawning to be sure.

Giulio Prisco April 16, 2016 at 0:41

Assuming molecular nanotechnology and mind uploading technology, we could print suitable robotic, biological or hybrid bodies on site, and then beam people to live in those bodies. There’s some excellent science fiction using this scenario, e.g. Greg Egan’s Schild’s Ladder and Incandescence.

But that’s long term. What I am suggesting here is much more conservative and could be a part of Starshot. I think the odds of developing real AI (like Samantha in the film Her) before Starshot launch (say 2050) are reasonably good. Then we should find ways to implement the AI in a StarChip with very low weight and power. If everything works, Starshot could carry a person to the stars.

A problem is boredom – what would the AI do after the fly-by?

Robert April 17, 2016 at 11:23

Molecular nanotechnology is at least based on proven scientific concepts. Mind uploading is so far basically a religious idea as is truly sentient AI. There is no scientific basis to think these could ever be done.

Giulio Prisco April 18, 2016 at 9:32

Robert, I don’t want to open a “religious” ;-) debate here, but I think sentient AI and uploading are both ideas compatible with science, so the burden of proof is on you. Why “there is no scientific basis to think these could ever be done”?

Robert April 18, 2016 at 13:10

That’s your opinion but you have any facts to suggest it is so? The concepts of sentient AI and mind uploading are mere speculation wrapped in scientific language. It takes a lot of faith to buy into such arguments and thus they are essentially religious ideas where the religion involved one might call Scientism.

Giulio Prisco April 19, 2016 at 7:31

You didn’t offer an argument, only restated your opinion.

There are plausible scientific arguments against the feasibility of conscious AI and uploading. While we are living proof that consciousness can arise in carbon circuits, perhaps it can’t arise in the silicon circuits that we know how to build. Perhaps some properties of carbon, for example quantum entanglement in biological brain circuits as proposed by Hameroff and Penrose, are really central to consciousness.

However, this argument doesn’t mean that machine consciousness and uploading are impossible. It only means that we will have to build suitable substrates that exhibit the key properties of carbon circuits. Harder, but doable.

Let’s agree to disagree. As always, time will tell.

Wojciech J April 15, 2016 at 10:40

Paul, I remember a proposal to use Sun’s gravity lens to communicate with interstellar craft.

Could we use this to communicate with the nanocraft swarm at Alpha Centauri ?


As to how such swarm might look like I am reminded of the quadrocopter formations that have been popularized recently:



Michael April 16, 2016 at 1:51

There is another possibility of using relays, if these light sails are sent out we can chose the exit velocity of them good enough to have a stream of them along a path. Communications are then sent down this pathway of relays. If in the far future we send large craft out we can then send these light sail communicators out later.

Project Studio April 15, 2016 at 21:50

“Hawking himself argued that intelligent machines based on mechanical and electronic components, rather than macromolecules, could eventually replace DNA based life, just as DNA may have replaced an earlier form of life.”

I’m pretty sure that was in the form of a warning, though. Hawking is not much of an advocate for AI.

Designing the processing final hardware and software would be premature at this stage – the approach to that needs to be flexible and agile as the engineering for other components such as the sail/mirror/transmitter and phased laser array/receivers matures. I’m sure the hardware and software will go through many revisions, and will take full advantage of the next twenty years of innovation and paradigms shift. For all we know, the OS might be quantum computing. Recent research has suggested diamonds would be a good substrate for the quantum q-bit storage. That should be fairly rugged.

I think that true AI as we imagine it is still a long way off, if ever. We still don’t know how human beings think and feel. Highly autonomous systems if programmed exceedingly well, will manifest a great deal of the intelligence of their human designers but are not likely to match it in real-life scenarios (of course those systems will process information extremely rapidly). Playing games, with defined rules and objects is one thing, but general machine intelligence in novel situations (as a star ship/chip will face) will not be possible until understand ourselves much better.

Giulio Prisco April 17, 2016 at 0:42

Hawking thinks of AI as both a huge opportunity and a huge risk, depending on how things work out.

“Success in creating AI would be the biggest event in human history, ‘ say Stehpen Hawking, Stuart Russell, Max Tegmark, and Frank Wilczek, in an article published on The Independent. “Unfortunately, it might also be the last, unless we learn how to avoid the risks… There are no fundamental limits to what can be achieved: there is no physical law precluding particles from being organised in ways that perform even more advanced computations than the arrangements of particles in human brains… All of us should ask ourselves what we can do now to improve the chances of reaping the benefits and avoiding the risks.”

Links, more quotes and personal thoughts here:

Re “I think that true AI as we imagine it is still a long way off, if ever. We still don’t know how human beings think and feel.”

That’s true, but do we need to know everything about how we think and feel to replicate the process? We can’t engineer a bird yet, but we can engineer a machine that flies much faster than a bird.

Of course, it may turn out that we can design super-smart thinking machines without a detailed knowledge of how our brains work, but we need such knowledge to design a machine that thinks and feels like us, sufficiently similar to a person to be considered as a person. Same applies to uploading.

If quantum effects are central to our type of consciousness, as some researchers think, then we need to engineer quantum substrates that exhibit the same quantum behavior, which would take more time, As always, time will tell.

Eniac April 24, 2016 at 8:30

In order to start the AI revolution, we only need to program an AI that can program an AI. Thinking and feeling like a human is a luxury, as is uploading.

Quantum effects have nothing to do with the working of our mind, as most researchers think.

Andrew Palfreyman April 17, 2016 at 20:16

The ISS is about 500 (metric of course) tons so we can launch 50 tons consisting of 5 gm starchips, over time no problem. That’s 10^7 starchips total. If we space them equally from here to Alpha Centauri, that’s an inter-chip distance of about 1 million Km. The relay comms idea should be doable for this toy example.

Michael April 18, 2016 at 13:01

We need not space them out to AC only to the SGFP’s which should reduce there number enormously, I believe a combination would be better.

Comments on this entry are closed.