Improving Safety in AI Research for Engineering Biology

AI News

The dangers posed by using data-centric methods to engineer biology have been identified by experts at the University of Bristol with the aim of making future research safer.

The potential misuse of data-centric approaches in synthetic biology poses significant risks: easier access to data science tools could allow bad actors to develop harmful biological agents for purposes such as bioterrorism, or to intentionally disrupt ecosystems.

The findings, published today in the journal Synthetic Biology, propose additional data hazard labels to describe data-related risks in the field of synthetic biology.

  • Accuracy of source data is uncertain – Because the accuracy of the underlying data is unknown, its use may produce erroneous or biased results.
  • Source data completeness is uncertain – The completeness of the underlying data is uncertain, leading to missing values ​​resulting in biased results.
  • Integration of incompatible data – Data from different types or sources are used together that may not be compatible with each other.
  • Potential to cause ecological harm – This technology, even if used correctly, has the potential to cause widespread ecological harm.
  • Potential experimental hazards – Safety precautions may be necessary to transfer the technique into experimental practice.

The research is the result of a collaboration between researchers at the Bristol Centre for Engineering Biology (BrisEngBio) and the Jean Golding Institute for Data-Intensive Research.

“We are in a transformative era where artificial intelligence and synthetic biology are converging to revolutionize bioengineering and accelerate the discovery of new compounds, from life-saving medicines to sustainable biofuels,” said Kieren Sharma, a co-author and doctoral student researching AI for cellular modeling in the School of Engineering, Mathematics and Technology.

“Our study uncovers potential risks associated with certain types of data being used to train modern systems biology models, including inconsistencies in measurements from complex, dynamic organisms and privacy concerns that could compromise the safety of next-generation models trained on human genomic data.”

This project is an extension of the work of the Data Hazards Project (, which aimed to create a clear terminology for potential dangers in data science research.

Nina Di Carra, PhD, from the School of Psychological Sciences, co-author and co-leader of the data hazards project, explained: “Having a clear terminology for hazards makes it easier for researchers to think proactively about the risks in their research and take steps to mitigate them. It also makes communication easier for people working across disciplines, who sometimes use different language when talking about the same issues.”

Interdisciplinary collaboration is essential to realizing these clear vocabularies.

Dr Daniel Lawson, Director of the Jean Golding Institute and Associate Professor of Data Science in the School of Mathematics, points out: “As datasets grow in size and ambition, increasingly sophisticated algorithms are being developed to gain new insights. This complexity makes a non-siloed, collaborative approach to identifying and preventing downstream harm essential.”

Dr. Thomas Gorochowski, associate professor of bioengineering in the School of Biological Sciences and lead author, added: “Data science will revolutionize the way we engineer biology to harness the unique capabilities of living organisms to address global challenges, from producing sustainable materials and fuels to developing innovative therapeutics. The extensions our team has developed can help bioengineers consider and discuss the risks associated with a data-centric approach to research, helping to ensure that the enormous benefits of bio-based solutions are realized in a safe way.”

This research was funded by the Royal Society, BBSRC and EPSRC and supported by the Bristol Biodesign Institute.


“Data Hazards in Synthetic Biology” by Natalie R Zelenka, Nina Di Cara, Kieren Sharma, Thomas E Gorochowski et al., Synthetic Biology

/Public Release. This material from the originating organization/author may be out of date and has been edited for clarity, style and length. Mirage.News does not take any organizational stance or position and all views, positions and conclusions expressed here are solely those of the authors. Read the full article here.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *