Dr Adam Stanton

Researcher in Adaptive Informatics

by as
Comments Off on New 3DRC work

New 3DRC work

I’ve been putting together a page that collates the work I’ve done on 3DRC, including some more recent analysis. It’s a work-in-progress and might remain so perennially, but I’ve made it available here anyway 🙂 It includes some more videos and plots that help to analyse the performance contributions of the various components. Hopefully this will culminate in another published paper, at which point I’ll also make the GitHub repo available for people who want to pick this up and explore it independently.

by as

Could we make ecobots?

A short proposal I made as part of a research fellowship application to an institute in Japan in Summer 2020.

Modern agriculture is dominated by industrial farming methods that are heavily and increasingly reliant on robotic automation. These methods destroy ecosystems and natural diversity, but are mostly unavoidable due to the need to feed the Earth’s growing human population. At the same time, there is a growing divide between people who embrace robotics in everyday life and those who think this is taking our species in the wrong direction.

My own perspective on the relationship between the human species and others is that we have sufficient awareness to recognise the intrinsic value of biodiversity as well as its very tangible benefit to us, and so we should consider its protection a defining goal of our existence. Another, more speculative idea that I would like to spend some time exploring tries to re-imagine the robot as a participant and supporter of natural systems, in order to change attitudes to robotics and food production. “Ecobots” are centimetre-scale robots whose function is twofold: they act as protectors of biological diversity and they also help humans to grow their own food locally, rather than by using industrial farming methods. “Locally” could include community gardens, private land, or even pots and planters in apartments.

The robots should develop an awareness of the members of the ecological community in which they are emplaced, learn to recognise the trophic relationships that define the interactions within this community, and discover the interventions they can make that maintain and strengthen these ecological systems, whilst at the same time providing an excess of food for human consumption.

This research project is very speculative since such robots are clearly a long way from practical implementation. However, in addition to making incremental technological progress, the attempt will constitute an opportunity to look at the problem of food security from a pan-species perspective (rather than the anthropocentric view we often find now), and also stimulate wider debate in society about the role of robotics and its relationship to the natural world.

Here, the dream is that humans re-imagine the robot’s relationship with nature and see such artifacts as contributing to a new era of biological unity on the Earth. Perhaps a part of this attitude is already more prevalent in some countries, but it will be interesting to see and compare reactions to it when framed this way in Japan and beyond.

PDF version

by as

How to use gscript to choose a random file from a Google Drive folder, and open it?

It’s not so hard, though the gscript has to be published as a web app. Write the doGet function to retrieve the URL of the file in Drive, then return a small script that redirects to the URL using HtmlService.

function doGet(e) {
  var folder = DriveApp.getFolderById("your-folder-id");
  var files = folder.getFiles();
  var filearray = new Array;
  var c=0;
  while(files.hasNext()) {
    var s = files.next();
    filearray[c] = s.getUrl();
  var item = Math.round( Math.random() * filearray.length );
  var itemurl = filearray[item];

  var html="<script>window.top.location.href='"+itemurl+"';</script>";
  return HtmlService.createHtmlOutput(html);

by as

What’s the Royal Institution like?

It’s amazing! I was invited to be part of a panel discussion on Artificial General Intelligence which took place yesterday. Quite a thrilling experience to be in the Royal Institution, talking about a subject I love, with extremely cool people: Wendy Hall, Raia Hadsell , and Brad Love! We covered some ground, bouncing between practical concerns about AI and a more sci-fi view of where we could end up, how we could get there, and what we might do to make sure AI is on our side.

The RI is a fascinating place to visit, let alone speak at. And I was told a great story about how the first ever “hack” happened there, to demonstrate the insecurity of Marconi’s wireless telegraph, in 1903.

by as

What does it mean to discover numbers?

I wrote an article recently that commented on research by Nasr et al (link) that showed a machine learning engine discovering a sense of number: AI develops human-like number sense – taking us a step closer to building machines with general intelligence.

This somehow led to a radio interview on the topic on BBC national radio–my first one, hence my slightly shaky start! My 98 year-old grandmother’s comment was “It’s all about apples!?” .. take from that what you will. Thanks a million to my good friend Kelcey Swain for chopping up the audio.

by as

What should we do?

Rather than just certain industries, let’s imagine that every job could, instead of by a human, be performed perfectly well by a machine.

Everything that humanity currently produces is still produced, except that people stand at the sidelines and observe rather than participate. From the outside, things work in the same way. One way or another, energy is still harvested from the universe, work is still done, and ultimately, the human species consumes the products of this process in the ways that it always has.

At this point, humans are at a tipping point. We are faced with two choices.

On the one hand, we could build machines to do the job of consumption as well as production, denying ourselves the last job available to us. This is the final nail in the coffin for the relevance of humanity as we know it. This scenario is the birth of Humanity 2.0: the machine version. Some of our essence, some of our ideals are carried into the future by intelligences unimaginable, and perhaps they will remember their history and celebrate our legacy as they decide where to take their reborn civilisation.

On the other hand, we provide the direction. Rising to the challenge of defending their relevancy, the children of today’s human beings unite behind a common endeavour. They ride out into the universe and sow the seeds of the human spirit across the galaxy, and take their creations along for the ride.

Automation on this scale is, though perhaps in the distant future according to a human lifespan, inevitable on timescales only an order of magnitude greater. Then if we take this scenario as a foregone conclusion, in either case it seems that the key question we have to address is one of purpose.

What is important on timescales of millennia? What should intelligent beings, human or otherwise, try to achieve, and what principles should guide their efforts?

In my view, our task is to raise human consciousness to a level where our ambition as a species (and thus our capacity to answer these questions) is, if not unified, then at least coherent.

How would our world change if our thinking were framed in these terms? Would we become conscious of the conspicuous lack of direction and narrow horizons in today’s discourse?

Over the horizon we can hear a warning roll of thunder: an automated world is coming. My hope is that we are inspired by its possibilities, that we realise that the immediate, practical problems that politicians and journalists wrestle with every day (including those that arise from the effects of increased automation in society) can be informed not just by an operational imperative for survival and the relief of suffering, but also by an empowering, strategic vision for human beings.