I trained a recursive neural network on old Southern Fried Science posts, then asked it about the future of ocean conservation.

Over the weekend, I decided to try my hand at some deep-learning using recurrent neural networks to create a text-writing bot trained on old Southern Fried Science posts. After 48 hours of training, the Southern Fried AI was born. This is what it has to say about the future of ocean conservation.


In the future, ocean conservation priorities will be dictated by the conservation conservation project. In the post of energy sharks and most of the greatest over the study has been conservation and dedicated to watch of conservation, the all new way that was she than a community to show better some astronation there is a research work (and sharks) in a more server shows of the research design of a discovery of campaign that even the results of some of the operational shark sources in the Climate change for the ensing and make that it may also end in the science of the response to the population of positions of marine scientists with my most months of cheap more and the project of a shark (with sampling and interesting proposal great from the final fishermen in the science and species of mermaids and other projects with some of the career of the Sea How and has a project that will be a dogfish down their beran place to go as a restoration and the ocean that submerse to have interesting a public posting ocean deciding enough to be asking “we could can be among the transports that is a shark critical components.”

Read More

Welcome to the Future: Three Rules for Artificially Intelligent Underwater Robots.

On January 1, 2016, the Southern Fried Science central server began uploading blog posts apparently circa 2041. Due to a related corruption of the contemporary database, we are, at this time, unable to remove these Field Notes from the Future or prevent the uploading of additional posts. Please enjoy this glimpse into the ocean future while we attempt to rectify the situation.


Underwater robotics has come a long way since I started working on it in the early ‘noughts. From the massive industrial beasts of the old guard to the small, sleak, eminently hackable sprite of the Connected Exploration movement to this new crop of fully autonomous, decision-making and directive setting AI-powered drones of the last few years, everything keeps getting smaller, cheaper, and more capable. It’s a great decade to be exploring the deep.

Last month, we deployed our first swarm of artificially intelligent deep diving robots designed to patrol the abyssal plane, identify regions of unique biodiversity, and recommend critical ecosystems for international protection in advance of biomining operations. What’s unique about this project is that we’ve assigned all decision-making authority directly to the swarm. They get to decide where in the world they go and how and when they sample. This came after years of debate and negotiation with stakeholders from science, conservation, and industry, and has been accepted through international agreement as the most unbiased and equitable solution to the challenge of getting groups with vastly different goals to agree upon dividing up the deep. Read More