If a cyberattack shut down the electricity grid, we would need to eat animal feed and biofuel crops, ban alcohol production, cut food waste and fish from sail boats. (Unsplash)The capacity of rogue AI agents to destroy our species within the next decade has recently been a hot topic of debate. Fears have been voiced within the industry about the rapidly increasing capabilities of artificial intelligence. If AI did choose (or was used) to eliminate humanity, long-term disruption of the electricity grid would be a straightforward approach. Analysis from the Council on Foreign Relations and the American Society of Mechanical Engineers outlines how sophisticated cyber operations could theoretically achieve this — a serious enough risk that Sam Altman, CEO of OpenAI, has been meeting with utility executives to discuss how to prevent it. Read more: AI agents can now remember and hackers can ‘poison’ their memories — a new cybersecurity threat An attack on the grid would hobble conventional agriculture. The only people with power would be those who generate their own by, for example, using solar power and microgrids. Without fuel pumps, lifestock ventilation, food refrigeration, irrigation systems, grain drying and fertilizer production, our food system would rapidly collapse. Even small-scale farmers would struggle. Without electricity, it would be challenging to irrigate and harvest large fields of arable crops. (Unsplash/Keagan Henman) We do not all have to starve, however. In fact, my colleague David Denkenberger and I wrote a book a decade ago that showed we could produce enough calories without conventional agriculture in a global catastrophe to feed the global population. Here, I share eight ways we can feed ourselves to survive if agriculture fails. How agriculture would failIf rogue AI agents took down the grid, conventional agriculture would fail in a predictable sequence. First we would run out of fuel. Pumps at fuel depots and stations are electric, so tractors, combines, irrigation generator sets and other equipment would stop working within days to weeks, depending on the extent of on-farm fuel storage. Next, livestock confined in controlled environments would die. Many poultry and swine barns rely on powered ventilation. Losses begin within hours of failure in hot weather and automated feed and water lines would fail in parallel. Food would also not last long without refrigerated storage, packing and transport. Milk, meat and produce spoil within days. Read more: ‘Agrivoltaics’ can both power AI data centres and increase food production — new study No irrigation, grain storage or fertilizerIn the medium term, the large amount of irrigation we use would fail as lift and centre pivot systems rely on electricity. Only gravity-fed and rain-fed irrigation systems would continue, cutting crop yields sharply in arid regions. Next up would be grain drying and storage failure. Wet-harvested corn would need to be dried and aerated with solar solutions. Without drying, mould growth would release mycotoxins, spoiling our stored grain reserves. The next growing season would see the end of synthetic nitrogen fertilizers. Produced through the electricity- and gas-intensive Haber-Bosch process, which converts atmospheric nitrogen to ammonia, this feeds around half of the global population. Without nitrogen resupply, cereal yields would fall roughly 30 to 50 per cent from depleted soils. No mills or slaughterhousesProcessing would then be the choke point. Many mills, slaughterhouses and packing plants are wholly electric. Cereals and livestock could exist in the fields and food would still be unavailable. To complicate everything, distribution and payment systems would also stop working. Rail signalling, trucking logistics and electronic payments would fail, fragmenting national food markets into local ones almost immediately. Small-scale farming communities would be in the best shape but even these would be challenged. The supply chains that have generated incredible increases in yields over the last century would break, returning us to yields from the early 1900s. Hybrid and genetically modified seed need to be repurchased annually. If only saved seed is available, it would segregate and yield would drop. Pesticides, veterinary drugs and equipment parts would also become much harder to obtain. Even communities that don’t rely on technology, such as the Amish, would be impacted, but conventional farmers would be in real trouble. This would affect everyone.Eight ways to feed ourselvesThe priority, of course, would be to pull the plug on the bad AI and fix the grid. In the meantime, reversion to manual and draft power would reduce output per worker by an order of magnitude and the global food supply would support a fraction of the current population. Subsistence and rain-fed agricultural regions such as Africa would fare better than industrialized ones such as North America.However, we do not all have to starve. Here are a few simple things we could do to feed ourselves:Stop feeding edible food to animals. Feeding food to animals and then subsequently eating those animals is inefficient. Redirecting grain from livestock to humans would be the single fastest calorie recovery we could achieve, available on Day One when the lights went out.Halt biofuel production. Edible crops that are currently converted to fuel would need to go back into the food supply immediately. Ban alcohol production. Crops used to make alcohol in the United States could feed 425 million people every year if we just ate them.Cut food waste. We should focus on leveraging existing calories rather than producing new ones. The Alliance to Feed the Earth in Disasters says conservation interventions are most effective in the initial stages of a global catastrophic food failure.Plant a high fraction of legumes to fix nitrogen. This will directly substitute for lost Haber-Bosch fertilizer and is the core agronomic pivot needed to keep fertilization working.Extract calories from agricultural residues. Straw and stover (leftover stalks and leaves) can be turned into animal feed and then into food via livestock. They can also be converted into sugars and then into edible products by microbes, used to grow fungi, used as a substrate for mushrooms or fermented into single-cell protein. Fish with wind-powered ships. Sail-based fishing removes fuel dependency entirely. We could haul in a lot of fish.Plant food in backyards. Do you have a lawn of grass you can’t eat? You could instead use it to make a lot of food and protein. Cultivating plants like soybeans in backyards could provide much of the protein households need. One easy way to feed ourselves would be to dig up decorative lawns and plant food in all backyards. (Unsplash) A last line of defenceIf things ever get really bad, my research group and others have developed more than a dozen ways to produce the calories that all of humanity needs while conventional agriculture is re-established. Our last line of defence would be creating so-called “resilient foods” to feed the population. These processes vary from converting plastic to food, producing an edible single cell protein from natural gas, extracting protein from leaves or even converting 3D printing waste to a single-cell protein feedstock that can be converted to human-edible food.These routes are more extreme — and hopefully unnecessary.Joshua M. Pearce has received funding for research from the Natural Sciences and Engineering Research Council of Canada, the Canada Foundation for Innovation, Mitacs, the U.S. Department of Energy and the Advanced Research Projects Agency-Energy, the U.S. Department of Defense, the Defense Advanced Research Projects Agency (DARPA), and the National Science Foundation (NSF). In addition, his past and present consulting work and research are funded by the United Nations, the National Academies of Science, Engineering and Medicine, many non-profits and for-profit companies in the energy and solar photovoltaic fields. He is a founding member of Agrivoltaics Canada. He has written extensively on GCR-related agriculture issues. He has no direct conflicts of interest.