By essence, GenAI instruments cannot handle directly this type of information. However, APIs contructed on top of the source datastores can be transferred to GenAI via the developpemnt of appropriate agents.
This is the core of present LLM (Large Language Models) based GenAI instruments. Having digitized an enormous amount of texts and determined their co-occurences they are able to correlate any submitted text (prompt) with their content and produce well-formed output texts. The latest concerns about AI Agents development points to their possible ability to produce formatted ouput which can be directly interpreted by other programs. Most popular output format being json.
=Knowledge comes from experience either directly acquired or obtained from reference past experiences It is common practice to structure knowledge (concepts) through thesaurus or semantic graphs in order to insure the reproducibility of references to the knowledge. The accumulation of experiences leads naturally to deductive logic instruments (expert systems). A question is to know or evaluate the capacity of GenAI tools to follow reasoning paths when they are provided with properly formatted information such as RDF encoded decision graphs
Models of Nature are essential when dealing with Earth Resources. Being either in the estimation of the amount of non- renewable resources to be extracted or in the evaluation of the sustainability of renewable resources regarding in particular long term evolution of climate. Models are also important in the evaluation of environmental impacts. Present GenAI tools are not equipped with these models but may be aware in certain circumstances of the interfaces to these models (ex: climate models), thus helping the automation of their use.
Any Earth Resources project has a social impact either locally or globally. The amount of information indexed by the GenAi tools can probably be exploited to extract characterisation and statistics on typical interactions between the development of projects and their social impacts. Behavioral simulation tools are also now being offered as a complement to GenAI.
All resources being either non-renewable or renewable need investments and running costs which need to be valuated and compared with their benefits either monetary or social. When dealing with non-renewable economics the term of "Full cycle economics" is used to evaluate the projects from start to exhaustion of the resource. In the case of renewable resources the economics is more organized around sustainability economic terms such as Life Cycle Costing and Sustainable Yield. In both situation it is expected that past experience accumulated can be usefully referenced using GenAI.
The evaluation of environmental impacts and the positive or negative social consequences is a crucial element in the decision of undertaking any Natural Resources project. This relates to the concept of Externalities evaluation. Past environmental and social data which should be available to GenAI tools could help in the evaluation of these terms.
Natural Resources projects meet with a score of objectives being either economical, environmental, social or even political. These objectives are often developed and promoted by very diverse groups of stakeholders which expertise and motivations are often conflictual. A proper governance of these conflicts necessitate that decisions must be properly mediated. This mediation can be conducted using multicriteria decision analysis tools. Although it is out of the question that GenAI tools could conduct such mediations themselves, the amount of accumulated multidisciplinary knowledge to which they access should provide a useful context to the understandingof stakeholders motivations.