Panel discussion ‘Are we switching on the turbo in biodiversity research with AI?’
The discussion following the presentations was lively and the audience was able to participate. On the podium (from left to right) were Dr Ralph S. Peters, Torsten Demuth, Dr Ameli Kirse, Prof Dr Chris Biemann and moderator Muschda Sherzada. © Roman Schwer
“Without AI, it would take us 1,500 years to discover and recognise all species,” extrapolated LIB insect researcher Ralph Peters. The consensus at the event “Are we switching on the turbo in biodiversity research with AI?” was that Artificial Intelligence – AI – can help us to better understand nature on various levels. The discussion at the “Hamburger Horizonte” event on Thursday at the Museum der Natur Hamburg centred on the possibilities and limits of what AI can achieve in biodiversity research.
“AI can recognise what is normal, what is common, but recognising what is new – and not knowing what might be new – is difficult,” said panellist Chris Biemann, Director of the Hub of Computing & Data Science at the University of Hamburg, describing the basic problem in taxonomic species recognition. This is why LIB taxonomist Ralph Peters sees AI as more of an assistant in species characterisation and human experts in a strong role, because: “The aim must be to know every species, even in its complexity.” He described how we lose many animal and plant species every day before we have learnt about them. However, without in-depth knowledge of species and their importance, there is no basis for conservation measures.
In contrast, LIB scientist Ameli Kirse saw many advantages in the application, in monitoring on agricultural land, in forests and meadows. For example, AI offers opportunities for greater efficiency, greater speed, cost savings and the formulation of new questions. The search for solutions for species protection in agriculture would therefore be significantly supported by AI.
Moderator Muschda Sherzada also brought the perspectives of LIB Centre Director Mariella Herberstein and citizen scientist Torsten Demuth into the discussion. ‘Without AI, we would not be able to use important information from the collections,’ emphasised Mariella Herberstein, referring to the 16 million objects in the LIB collection.
The panellists summarised in statements the extent to which AI can give biodiversity research a boost:
Prof. Dr Mariella Herberstein, Head of the Centre for Taxonomy and Morphology at the LIB
“In order to really understand biodiversity, we need to link so much different and rich data that without the support of artificial intelligence, we will get through this forest of data far too slowly. Time is running out, and if we need to speed up biodiversity research now, then we can only make progress with the help that artificial intelligence, for example, can provide.”
Prof Dr Chris Biemann, Director of the Hub of Computing & Data Science and Professor of Computer Science for Language Technology at the University of Hamburg
“AI can support biodiversity research in all areas. It starts with the collection of data, for example with automatic photo traps for counting insects. During analysis, AI models can help to identify and evaluate patterns and trends more quickly and to organise the data faster. And finally, AI can also be used to make zoological collections more accessible to laypeople in order to communicate the importance of biodiversity to society at large.”
Dr Ralph Peters, Head of the Hymenoptera Section at the LIB
The majority of the world’s species are still unknown. AI is helping to fulfil the mammoth task of discovering biodiversity. For us, AI is an assistance system that only works in conjunction with humans, who supervise, check and further develop the AI and its results. By taking over repetitive and formalised tasks that are rather tedious for humans, AI will not only make biodiversity discovery faster, but also more fun. Rapid biodiversity discovery is essential in the face of massive biodiversity loss.
Dr Ameli Kirse, research associate in the AMMOD/BioMonitor4CAP project at the LIB
‘Artificial intelligence has the potential to significantly advance and accelerate biodiversity research. Through the automated analysis of large amounts of data, it enables the rapid identification of species using image and sound analyses and recognises patterns in ecological data, which allows timely decision support for the protection of endangered ecosystems. In addition, AI contributes to the more efficient use of resources in field studies. However, the accuracy of the results remains a major challenge.’
Torsten Demuth, citizen scientist, Verein Neuntöter e. V.
“There is still great potential to speed up work with collection material, especially when it comes to image and pattern recognition. The models that are currently freely accessible are sometimes impressive, but they are also regularly sobering. In the area of fieldwork, I see little practical benefit in the medium term; only small sub-areas are currently covered here. All in all, the quick successes are deceptive, and there are still many unanswered questions regarding the political course.”
This year’s Hamburg Insect Week kicked off with the discussion event. Further topics on biodiversity protection in the Hamburg Insect Week: www.insektenwoche.de
