Jonathon Welch is the Head of Data Science at Albert Invent. Bringing a wealth of knowledge and research experience to his role, Jonathan is leading the continued innovation of Albert’s end-to-end R&D platform which includes Artificial Intelligence (AI), Machine Learning (ML) and more.
There has been a whirlwind of discussions about AI, its future and its place in society. Most recently, due to an open letter from industry leaders, who urged a six month pause on new AI development to enable the industry to instead focus on what is needed to make AI more accurate and safer to use.
In this discussion with Albert’s head of Data Science, Jonathan Welch, he shares his perspectives surrounding AI. Jonathan brings both a passion and in-depth experience in data science, including a Ph.D. in Applied Physics from Harvard.
Listen to AI discussion with Albert’s Head of Data Science, Jonathan Welch or continue reading a high-level summary of his perspective below!
What are some of the overall concerns with AI?
Anytime there is a transformative technology, that creates a paradigm shift in how we think, it comes with growing pains. While there are great opportunities with AI, we’re also seeing an acceleration of use-cases that are creating uncertainties about its impact it may have on our society. This is what caused leaders more recently to ask for a pause to give researchers and experts time to develop some guidelines and guardrails around AI development, use, and place in our society.
There are of course the overarching concerns around misinformation and sharing of unvalidated data, though two key concerns are also front and center – the potential impact on critical thinking and loss of jobs.
Potential impact on critical thinking
Using AIs to solve things like academic tasks or homework or essay writing is bringing up concerns about the impact this could have on the educational process. As learning algebra, doing a literary analysis or building and supporting an argument with evidence teaches our brains how to think. If we take away those academic muscle builders and rely on AI to solve those problems for us, it will be more difficult to develop critical thinking skills.
With AIs often functioning as a question-and-answer machine — this could lead to what is sometimes referred to as the “black box” effect — where you don’t necessarily know how the answer is generated or what it is based on and lose engagement in the decision-making process itself.
Potential impact on jobs and the world of work
Anytime there is a transformative technology, there is always the concern that it will replace jobs, and AI is no different. Though for the potential job loss, there are also new jobs that are and will be developed because of AI. As perspective, while AI can automate certain tasks and decisions, it’s also important to note that by doing so, it also frees up time and mental energy for people to focus on other areas that require creativity, critical thinking, and problem-solving skills.
What about AI use in our industry – chemistry and materials science?
At Albert, we are focused on leveraging AI to expand rapid innovation, empowering scientists with new capabilities so they can explore new ideas and achieve results faster than before. As a result, we aren’t building AIs to replace humans, but instead on those that enhance the capabilities of individuals in our industry.
An example of this is facilitating collaboration and knowledge transfer. By having an AI that is trained upon data that is collected by your entire organization, it provides the ability to transfer knowledge from domains you may never as an individual scientist have ever thought to look into, serve up trends, or learn about correlations that are useful in your own work drawn from your colleagues’ work. This is just one of many use-cases of AI and the positive impact it can have in expediting research and development.
The underlying question that impacts the use of AI in our industry is how well we can trust the data that the model is trained on. Any inherent bias in the data can lead to some form of misinformation in what is delivered by the AI. This can be particularly problematic in the field of chemistry and materials sciences and that is an important thing to consider when bringing an AI tool in your workflow.
Key to the successful use of AI in our industry
To ensure the success of AI, it’s crucial to maintain a clean and well-structured data ecosystem. The quality of an AI model heavily relies on the quality of the data it’s trained on. Therefore, it’s important to establish clear and rigorous policies regarding data validation, cleanliness, and lineage. These policies should apply not only to internally generated information but also to any data sets that come from external sources into your centralized source of truth. By adhering to these policies and practices, organizations can have confidence in the reliability of their data ecosystem, which is crucial for extracting meaningful insights from any model built on that data.
Bottomline: AI is not going away
AI presents both challenges and opportunities for our society and industry. To fully leverage its potential, it’s crucial to determine the best ways to use it, especially in our industry context, to achieve successful outcomes.
To ensure that AI is set up for success, we need to lay the groundwork by providing it with clean, trustworthy, and validated data. As long as we do that, we really shouldn’t shy away from AI usage, but we should learn how to integrate it in a way that enhances our ability to innovate faster.
Listen to this informative discussion with Albert’s Head of Data Science, Jonathan Welch, for additional perspectives on Artificial Intelligence.