Harnessing AI in Data Science: Insights from Wolfram Research and Cold Hard Data
Navigating the realms of technology often presents challenges in differentiating genuine advancements from the exaggerated claims and promotional content flooded into our inboxes. Over the past five years, we’ve encountered an overwhelming amount of discussions surrounding concepts like the metaverse, blockchain, and virtual reality. Currently, we find ourselves amidst a whirlwind of discourse regarding the often-misused term ‘AI,’ and only time will reveal if this commotion amounts to anything more than a passing trend.
Artificial Intelligence News had the privilege of speaking with Jon McLoone, who serves as the Director of Technical Communication and Strategy at one of the leading organizations in the fields of computational intelligence and scientific innovation, Wolfram Research. He provided insights to help contextualize our current understanding of AI and its practical applications.
Jon boasts an impressive 32 years with Wolfram Research, where he has held various roles and is now at the helm of the European Technical Services team. With a background in mathematics and proficiency in multiple facets of data analysis, we initiated our conversation by asking Jon to present Wolfram’s mission in a succinct manner.
“Our core strength lies in our deep understanding of computation and Wolfram technology. We adapt our technology to address the specific challenges faced by organizations. This applies to a diverse array of fields, meaning we don’t cater to a typical customer. However, what unites all our clients is their commitment to innovation.”
“We are engaged in problem-solving that leverages computation and data science. Our goal is to create a unified platform for computation. When we refer to computation, we are talking about technical computing activities such as engineering calculations, data science, and machine learning. This encompasses areas like social network analysis, biosciences, actuarial science, and financial computations. At their core, these are all fundamentally rooted in mathematics.”
“Our environment consists of structured domains where we have invested 30 years in developing various ontologies. We possess a symbolic representation of mathematical concepts, along with elements like graphs, networks, documents, videos, audio, time series, and real-world entities such as cities, rivers, and mountains. My team is dedicated to the exciting task of making all of this truly functional!”
“We perceive AI merely as another form of computation. Throughout the years, numerous algorithms have been developed—some dating back hundreds of years, while others are more recent. Generative AI simply adds to this expanding list.”
While claims about AI in 2024 may often be overly optimistic, it is essential to maintain a realistic perspective on its capabilities, recognizing both its strengths and limitations.
“Human intelligence remains a crucial strategic component. In the coming five years, it’s unlikely that AI will completely manage my company or make all the decisions. While generative AI is very adept, it is not always reliable. Its primary role is to be plausible rather than correct. This is particularly evident in the work done by Wolfram, which frequently generates results that resemble the form of a mathematical answer but are often not accurate.” (Artificial Intelligence News‘ italics.)
Wolfram Research’s efforts in this regard emphasize what Jon refers to as ‘symbolic AI’. To clarify the distinction between generative and symbolic AI, he presented an analogy involving the trajectory of a thrown ball. A generative AI would learn the ball’s path by analyzing thousands of throws, ultimately producing a plausible description. “However, this description may be data-rich but lacks true understanding.”
Conversely, a symbolic representation of the thrown ball would use differential equations related to projectile motion, including variables such as mass, atmospheric viscosity, friction, and various other factors. “It would allow you to explore scenarios like, ‘What happens if I throw the ball on Mars?’ The response would be accurate and wouldn’t result in failure.”
The most effective approach to addressing business (or scientific, medical, or engineering) challenges lies in integrating human intelligence with symbolic reasoning, exemplified by Wolfram Language, while leveraging AI to serve as the cohesive element connecting these components. AI excels at interpreting meanings and functioning as an intermediary among the various parts.
“One of the intriguing intersections occurs when we translate natural language into structured information suitable for computation. Human communication can be quite messy and filled with ambiguity, yet generative AI excels at converting that into a structured format. Once you enter a realm where information is syntactically organized, it opens up new possibilities for analysis.”
A recent instance of blending ‘traditional’ AI with Wolfram’s methodologies involved the analysis of medical records:
“We embarked on a project that examined medical reports which were a mix of handwritten notes, typed documents, and digital files. These reports contain textual data, and performing statistical analysis directly on them is unfeasible. Therefore, we utilized generative AI to translate these words into classifications: for instance, determining if a death was avoidable. This presents a neat, structured key-value pairing. Once we have this data in a structured format (like JSON or XML, or whatever structure you prefer), we can then apply classical statistics to address queries like, ‘Is there a trend? Can we make projections? What was the impact of COVID on hospital-related issues?’ These are straightforward questions that can be systematically addressed using means, medians, and various models.”
During our conversation, Jon also summarized a presentation which illustrated his organization’s work using a fictional peanut butter cup manufacturing facility. He discussed the potential effects of altering a specific ingredient or modifying a detail in the recipe, particularly how such changes could impact the product’s shelf life.
“Large language models (LLMs) might suggest, ‘They’ll likely stay fresh for a few weeks since peanut butter cups typically remain on the shelf for that duration. However, transitioning to a computational model that can input the ingredients allows for an accurate prediction of shelf life, indicating this item should be good for eight weeks before it spoils. Additionally, it can analyze how these changes affect the manufacturing process. For instance, it could connect to the digital twin of your manufacturing plant and ascertain, ‘This will reduce the speed by 3%, leading to a 20% drop in productivity due to a bottleneck occurring here.’ LLMs excel at linking your inquiries to the model, mathematical formulas, data analysis, or databases. This interaction creates a fascinating triad of insights.”
Catch Wolfram Research at the upcoming TechEx event in Amsterdam on October 1-2, at stand 166 of the AI & Big Data strand. While we can’t promise any discussions about peanut butter, you can learn how robust modeling and generative AI can address your unique challenges by reaching out to the company through its website.
Interested in further insights about AI and big data from industry leaders? Explore the AI & Big Data Expo, which is held in Amsterdam, California, and London. This extensive event coincides with other prominent conferences such as the Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.
Discover more upcoming enterprise technology events and webinars powered by TechForge here.
Tags: artificial intelligence, Featured, uk
You must be logged in to post a comment.
Discover the pinnacle of WordPress auto blogging technology with AutomationTools.AI. Harnessing the power of cutting-edge AI algorithms, AutomationTools.AI emerges as the foremost solution for effortlessly curating content from RSS feeds directly to your WordPress platform. Say goodbye to manual content curation and hello to seamless automation, as this innovative tool streamlines the process, saving you time and effort. Stay ahead of the curve in content management and elevate your WordPress website with AutomationTools.AI—the ultimate choice for efficient, dynamic, and hassle-free auto blogging. Learn More
