Foresight 8. Precision fishing

Background

In the sci-fi franchise Terminator, the malevolent super-intelligence known as Skynet achieved sentience or awareness on 29 August 1997. The event was the culmination of the fictional Cyberdyne Systems trying to remove human error and the slow response time associated with decision-making under a nuclear attack. In the real world, these reasons are the prime motivation for high frequency, algorithmic trading on sharemarkets, which has itself resulted in alarming events. As a result it is fair to say that both popular fiction, and real-world trends, have instilled a general fear of AI technology in society, and even in successful tech advocates.

The thesis of this foresight piece is that we are headed for a Skynet Future, but of a more benign, or at least a less menacing nature – in which we will see fishing take advantage of AI.

We have noted the impending disruptive effect of current AI tech, i.e. AI without sentience or awareness, and a great deal has been written on the subject. In addition to the advancement in software capacity of AI tech however, other innovations will likely interact to realise a Skynet Future, and thus compound the effects on policy, science and society, in general, and on oceans, in particular. These innovations include:

1. Sensorisation of the built and natural world (increased observation capacity)
2. Increasingly networked small microprocessors, e.g. single board computers; SBCs (increased hardware processing capacity)
3. Integrated communications technology (increased communication capacity)

The combination of these trends into devices that collect and transmit data through the internet has been termed the Internet of Things (IoT). In 2008, there were more devices connected to the internet than people. Today there are certainly more, and the trend is expected to continue. Driver-assisted cars are currently operating. By 2020 autonomous cars will start operating on roads that don’t have pedestrians. Fully autonomous cars will be operating on roads within 15 years (2035) the average automobile fleet turns over rate. Insurance premiums will complete the effort, and so by 2050 human drivers will be the exception.

Scenario

Precision farm management is a concept based on observation, measurement and response to crop variability, with an aim of defining a decision support system (DSS) to optimise financial returns. Often it is associated with advising planting schedules, based on soil maps, and remote sensing data, wireless networks, and possibly even robotics. Usually, decisions are based on DSSs, but the final decisions remain with the farmer, or farm manager. The concept of precision agriculture is easily transferable to fishing. The relatively short-term process of catching fish, which does not rely on the length of a growing season, however lends real-time or near-real-time operations, particularly in terms of price and economic data, current environmental (oceanic) conditions, all transmitted through phone or satellite communication network. In addition, with recent understandings on the limits, biases and constraints of human cognitive capacity in decision-making, there is potential for greater margins to be made by removing humans from the process, with an AI. Thus, observation, communication, data processing and decision-making, would result all without the need for human intervention.

Under such high profit motivation we might expect to see a greater move to privatise the fishing commons, and move to catch shares and ITQ management of fisheries.

1. Number of fisheries managed by ITQs, catch shares or IQEs. (See also project paper Little, The Rise of DAOs and the Decline of the High Seas Fishing Commons).

With the increased focus on profit margins for companies and industries, precision fishing will affect the entire workflow of the fishing industry from operation planning; fishing operations; sales and processing; and management.

1. Fishing operations planning: Planning fishing operations will rely on forecasting ability: market price forecasting, environmental ocean forecasting in addition to biological forecasting, augmented with propriety personal (historical fishing) knowledge. Decisions will be made by an AI system to optimise CPUE, and minimize operating costs.
2. Fishing operations: It is likely that sensorised gear will monitor the state of the environment / ocean, to improve catchability.
3. Sales and processing: Network connectivity will likely provide just-in-time (JIT) supply chain planning for markets and processors.

Fisheries management: The development of real-time (RT) of near-real time (NRT) data acquisition would enable stock status estimation and management on a similar spatial scale.

Indicators: How would we know this is starting to happen?

1. Number of fisheries managed by ITQs, catch shares or IQEs reaches 80% in Australia

2. Nominal fishing effort (fishing power increases). Number of vessels declines by 25%.
3. Nominal wild capture in Australia will not decrease but maintain between 0-20%, 170,000t to 200,000t (ABARES)
4. Employment in the wild capture fishing industry declines 33% (from current 5600 in 2016 to 4000 – ABARES)
5. Number of real-time (RT) of near-real time (NRT) fisheries with data acquisition or assessments will exceed 50%.

Scoring of indicators

Project team only – “score” this scenario (requires login): Click to continue.

Additional reading

Christiani, P., J. Claes, E. Sandnes and A. Stevens (2019). Precision fisheries: Navigating a sea of troubles with advanced analytics Advanced analytics may help struggling fisheries thrive while simultaneously protecting endangered ocean resources. McKinsey and Company.