Page 45 - Vol 32 Issue 33 2121

artificial intelligence mining

AI in Mining: 6 things you

ought to know

By Leanne Mostert, a Partner at Webber Wentzel

prevailing data tested before use. Their use should be
protection and appropriately supervised, including
confidentiality ensuring circuit breakers are in place if the
l a w s . behaviour exceeds certain boundaries.
Consider the Ethical use
consequences
of third parties Mining companies should consider the
acce ssing future impact of AI on the business. How
the data and does it fit into the broader corporate
ensure there social responsibility policy? Is it necessary
is either a data to have an ethics committee or another
processing or board committee to oversee AI? What
data sharing is its impact on the workforce? It is also
agr eemen t important to be transparent about the
in place. If use of AI, where possible, and to track
rtificial intelligence (AI) offers possible, anonymise the data (remove government and regulatory responses to
AI.
prospecting and mining possible identifiers in personal data) to
Aoperations a range of new avoid these concerns. Use regulatory IP
possibilities to enhance efficiencies. or development sandboxes, which are
isolated environments in which testing AI raises various challenges for patent
These include a greater ability to target can be conducted without harming the law, particularly whether the existing
likely mineral deposits, plan new mine host network. Where necessary, conduct legal framework can accommodate the
developments in difficult terrain, a data protection impact assessment. intelligent machine.
control autonomous vehicles and The first barrier relates to patenting the
improve recovery rates from diamond Liability and regulation original algorithm. In most countries
sorting or other minerals processing. AI may make decisions for which the (including South Africa) abstract
As AI applications are developed, mining company will be held responsible. mathematical methods are not per
there are at least six important legal If the AI system is provided under a se patentable. The second barrier is
aspects mining companies should contract, the contract must address the whether the outputs of AI systems can
consider in order to avoid future standards required and include limitation be patented, particularly where the
embarrassing and costly disputes. and exclusion provisions. If the system is AI system is independently inventive
Collaboration/ownership not provided under a contract, the mining without human involvement. The key
company needs to consider its liability
If the AI project is being undertaken in in the event of infringement of rights unresolved issue here is whether a human
inventor is required for an invention to be
collaboration with another party, mines and whether an appropriate disclaimer patentable.
should establish who will “own” the would be necessary. When the system
outcome, taking into account issues takes decisions about individuals, it is We expect the view of patent offices
such as the value that each party brings, important to ensure that the process globally will be that a human inventor
the elements that the mining company is fair, lawful and not discriminatory. A is required. If so, then inventions by AI
is most interested in (algorithm, data, system that takes important decisions systems may not be patentable under
output of algorithm) and the commercial about individuals requires additional the existing legal framework. Protecting
aims for each element. The appropriate controls, such as informing them and AI outputs is currently challenging the IP
intellectual property (IP) agreements allowing them the option of a human legal framework.
should be in place, as well as other re-evaluation. The system should also be
contractual protections, such as rights to secured from cyber-attacks.
use, exclusivity and confidentiality. Safe use
Developing AI & data
AI systems could behave unpredictably.
It is important to know whether the use Their potential dangers should be factored
of data in training and testing the AI into the mining company’s broad risk
project continues to be compatible with framework and they should be properly Leanne Mostert

AFRICAN POWER Mining & Oil Review Vol 28, Issue 29, 2019 | 45