b'Canberra observed Canberra observedtensor-flow deep neural networks andare now approximately 20% of what many more. It will obviously be somethey were between 2013 and 2015 time before AI successfully identifies(Figure 1). It was, therefore, good to drill sites for new mineral deposits, butsee that the Government recently if these machines can beat humans atreleased 64 areas for offshore oil and Chess and Go we can expect greater usegas exploration. Bidding is open until of AI as an exploration tool in the near5 March 2020 and this release should future. provide a welcome boost to our Explorers have to face the reality that,exploration effort.in Australia, most of the easy-to-find- The release areas are in five basins: deposits have already been found. ItsBonaparte Basin (15), Browse Basin (3), just going to get harder to discoverNorthern Carnarvon Basin (39), Otway more resources and we are going toBasin (2) and Gippsland Basin (5). Figure 2 David Denham AMneed new ideas and better skills toshows their locations.Associate Editor for Governmentdo it. Conferences such as AEGC 2019 denham1@iinet.net.au have a crucial role in that regard. They provide opportunities to network,Minerals exploration investment collaborate and generate ideas. And thatcontinues to increaseAEGC 2019 a greatis how conferences such as this one will Meanwhile, the rebound of the minerals success, but how willcontribute to discoveries yet to be made.sector shows no sign of slowing down. it contribute to moreBut it is not going to be easy. what it was in the 2016 March quarter. Investment is now more than double discoveries? Petroleum exploration investmentThe drilling numbers (Figure 3) are a stagnates useful indicator of the health of the With more than 1000 delegates fromsector, and with over 2.5 km drilled in the over 20 countries, the Perth conferenceThe petroleum sector shows no signJune quarter it is clear that the upward was a big success. There were over 400of recovery, and investment levelstrend is continuing.contributions as either poster or oral presentations, and the subject scope was huge. It ranged from a detailed seismic survey to locate an old gun emplacement, through to continent-wide EM, magnetic, gravity and radiometric surveys and their interpretations, to the 872 km land seismic line from Kiwirrkurra to Marble Bar over the Kidson Basin.You can only attend a few of the papers presented, but I was impressed by two that analysed the evolution and prospectivity of the Bight Basin. Using gravity and seismic information they derived models of the tectonic history, structure and morphology of a 600 km cross-section. The interpretation covered the time window from when Antarctica broke away from the Australian continent about 100 Ma to the present-day. A masterly display of extracting the most from the available geophysical data.As expected, AI and machine learning techniques were more prominent than in the 2018 meeting. There were presentations on 3D stochastic geological modelling; machine learning techniques on airborne geophysical data for mineral resources exploration;Figure 1.Quarterly petroleum and mineral exploration investment 2005-2019. Investment in minerals is now mineralisation predictive targeting usinggradually increasing, but petroleum expenditure is stagnating. Data from the Australian Bureau of Statistics.OCTOBER 2019 PREVIEW 34'