All About Dual Use Technologies
If technology is inherently dangerous, what makes AI so different?
A common phrase closely tied to policy discourse surrounding artificial intelligence (AI) regulation is its categorisation as a dual use technology. This phrase can refer to any technology, equipment, or knowledge that has both civilian and military applications. Simply put, some new innovations or technologies can be utilised for peaceful purposes including scientific research, industrial production, or healthcare delivery, but may also carry the potential for being repurposed towards military applications, such as weapons development.
One of the earliest explicit attempt to regulate a ‘dual use’ technology was the initiative of the Nuclear Suppliers Group (NSG), which was established in 1974 to control the export of nuclear material by supplier countries. While civilian applications of nuclear as a ‘relatively clean’ source of energy were appreciated, the transfer of nuclear fuel or fissile material carried major risks, if it were to fall into the wrong hands and used for enrichment to create a nuclear bomb. The NSG and sought to establish guidelines for nuclear exports to prevent nuclear weapons proliferation, while enabling civilian power-generation using this technology.
In the nuclear context, restrictions under the NSG guidelines (INFCIRC/254, Part 2) refer to:
those technologies (whether hardware, materials or software) which could make a major contribution to a nuclear explosive activity, an unsafeguarded nuclear fuel-cycle activity or acts of nuclear terrorism.
As nuclear energy, when introduced, one of the most destructive technologies known to mankind by an order of magnitude, the ambit of restricted dual use items that could fall under the specified guideline categories is notably broad. It becomes very much apparent that the destructive context of nuclear technology has dictated the extent of export restrictions, and not merely their dual use nature alone.
Nuclear energy now comprises 10% of global electricity production and has become the second largest source of low carbon power in the world. While nuclear energy has continued to provide reliable electricity to millions of homes in specific countries, it hasn’t managed to achieve anywhere close to its global potential. This is in part due to the complications introduced by the NSG, in procuring the relevant technology and nuclear materials. India itself only managed to arrange supply of nuclear fuel from NSG member countries at great political cost.
Separately, the NSG guidelines have not been entirely successful in preventing proliferation. The purpose of highlighting these ramifications is not to criticise the NSG, but to point the log-term implications that these kinds of policy choices can carry.
Fertiliser or Bomb?
Travelling back further back in time to the start of the 20th century, we can see also see that the principle underlying the dual use conundrum is not a recent development. For instance, the seemingly innocuous discovery of ammonia, a compound vital for agriculture and food production revolutionised a number of industrial sectors. While it has primarily been used as a fertilizer (about 80% of the production of ammonium nitrate is used for this), ammonia can quite easily be employed in the production of explosives. Many of us recently witnessed this destructive potential first hand through the horrific explosion of 2,750 tonnes of ammonium nitrate in Beirut on 4th August 2020. While this particular incident was accidental, the same compound was used in major acts of terrorism around the world, stretching from Brussels to New York.
At the moment, there is hardly an attempt being made to reduce ammonium production despite its well-acknowledged harmful potential. The world has fundamentally determined that the benefits of the chemical clearly outweigh the harms of its misuse. Ammonia is not seen as an existential threat to humanity. However, the dual use example of ammonia underscores the perennial challenge of managing technologies that possess both benign and destructive potentials.
Why does it matter?
The implications of being categorised as a dual use technology can be quite severe when it comes to international regulations and export controls. As the world is grappling with the onset of the ‘age of AI’, the dual use nature of this technology is informing calls for its regulation through an internationally acceptable framework that focuses on restricting its negative applications. The very same restrictions may inhibit the positive socio-economic changes which a AI may introduce into India, and man other countries.
AI’s application can potentially revolutionise healthcare, automate public transportation, or enhance industrial productivity. At the same time, countless writings have made us well aware of the more sinister uses of this technology, including the creation of AI-catalysed deepfake content, automated cyberattacks, and facial-recognition driven surveillance systems. However, the mere categorisation of any technology as dual use should not be a central reason for domestic strict regulation, or export controls. There are certainly other risk factors, along with the assessment of positive or harmful outcomes of different AI use-cases, which ought to become a part of this conversation.
The central question should be: Is AI just another entry in the long history of dual use technologies, or does its unique potential necessitate new approaches to governance and responsibility?
I am an independent Delhi-based lawyer working with technology, law and policy. The views expressed in this post reflect my personal opinion.