Beyond the Hype: Providing Computational Superpowers for Enterprise AI

Whether you’ve noticed or not, artificial intelligence is becoming integral in modern systems, either front and center or behind the scenes. And this is just the beginning. Thanks to recent advances in large language models (LLMs), examples range from customer service chatbots to health-care data analysis and nuanced writing advice. This era has arrived with conversational interfaces, the processing of unstructured data, code synthesis and simple cognitive tasks.

Beneath this veneer of sophistication, however, lies a critical reality: LLMs are not a panacea for all computing challenges, especially given their tendency to produce results that are plausible without necessarily being accurate.

(That is, as Carnegie Mellon University professor Jignesh Patel put it: “Generative AI exceeded our expectations until we needed it to be dependable, not just amusing.”)

And if you need LLMs to make use of your enterprise data, models or algorithms, this is a very big issue.

Fine-Tuning Limitations

Fine-tuning LLMs was initially seen as the solution to inaccurate answers and hallucinations because it allowed models to be adapted specifically to particular domains or tasks. By exposing the LLM to a curated set of domain-specific data, the model could learn the nuances and specialized knowledge required to generate more accurate and contextually relevant responses. This process promised a significant reduction in errors and improved performance in niche applications, making it an appealing approach for early adopters.

Fine-tuning LLMs to increase accuracy and reduce hallucinations ultimately revealed a significant limitation: fine-tuned models tend to become rigid and less adaptable, struggling to incorporate new information or contexts without additional retraining, which is impractical in rapidly evolving fields. This challenge highlights the need for more flexible and scalable approaches to improving LLM performance.

The Retrieval-Interleaved Generation (RIG) Paradigm to the Rescue

Fortunately, the retrieval-interleaved generation (RIG) paradigm addresses many of these limitations. Instead of relying solely on the static knowledge embedded in the model (that is, from its training data), the LLM is connected to external sources such as databases, knowledge systems or even the web. When it encounters a query that requires current or domain-specific information, the model retrieves relevant data dynamically and incorporates it into its generated responses.

This was the reason Wolfram was invited to be among the first plugins to ChatGPT (something that has since evolved into Wolfram GPT). That plugin used Wolfram|Alpha as a source of data and the Wolfram Cloud as an engine for executing Wolfram Language code that the LLM might synthesize.

This becomes relevant for most enterprise applications accessing private, proprietary data, whether that is billing and shipping data for a customer services chatbot; production, stock and orders data for a manufacturing control tool; or scientific and engineering models for a research assistant.

Of course, within Wolfram Language, we have a well-established pipeline of technology that makes it trivial to connect to various data sources and add computational tools to an LLM that can do all of that, so for simple projects, this is already a solved problem.

The Scaling Challenge

Unfortunately, while the approach of “take an LLM, add some prompt engineering and add some tools” can quickly make great applications for narrow purposes, it can start to break down as you broaden the aspirations for your tool. The problem? As you add more endpoints to each of your different databases or for multiple models and digital twins, this level of complexity can overwhelm your LLM and cause it to be confused.

For example, a financial analysis tool leverages multiple databases for various data types, such as stock market data, economic indicators and company financials. When a user asks for insights on how a recent economic indicator change might affect the stock market, the LLM needs to fetch data from both the economic indicators database and historical stock market data to analyze correlations. The LLM might, however, mistakenly call the stock market database for economic indicators or vice versa or send incorrect arguments like date ranges or specific indicators to each endpoint. This can result in the tool providing inaccurate or incomplete information, frustrating the user and diminishing the tool’s reliability.

The problem is twofold. First, the LLM starts to get confused about which endpoint to call for which piece of information and the arguments to send to the endpoint. But more profoundly, when you ask queries that cross different silos—say joining data or passing retrieved data into a model to produce a prediction—it gets confused about what things really mean. This is as much a feature of the ambiguity of human language, which is the LLM world, as a problem with the LLM. (It is, after all, why math and other forms of symbolic representation and processing were invented.)

The Computable Knowledge Layer

One solution to the scaling challenge is to produce an all-encompassing endpoint that is a single source of computational knowledge and data and where all these issues of symbolic meaning, source identification, formal representation and processing are taken care of. You then provide a single, flexible interface that the LLM can send its knowledge queries to.

Sure, the LLM still has to call this endpoint correctly, but Wolfram has already mastered this type of challenge thanks—once again—to our earlier work with Wolfram|Alpha. It’s a knowledge engine designed to be a single source of computable data—albeit, originally for direct human access—from private knowledge sources and ontologies. Furthermore, it also has a natural language interface that, while far less fluent than modern LLM approaches, is nevertheless sufficiently forgiving and broad for the LLM to communicate with natural language, which it naturally does, without having to try and teach it to use formal API codes.

Making Your Data Computable

So what is involved in getting data ready for LLM access? At a small scale, nothing. If you have relatively narrow goal and clean data sources, you can deal with the challenges through a combination of endpoint design and prompt engineering. Indeed, we are engaged in several “add an LLM to my data” type projects from database or document sources built directly with combinations of Wolfram Language LLM-related functionality, Wolfram Chat Notebooks and deployment technologies like Wolfram Enterprise Private Cloud.

But while you are getting these “easy wins” in place, you should start considering preparing your data for the more ambitious “make my entire enterprise knowledge accessible to AI” type projects that will soon become one of the decisive competitive advantages for many organizations. This requires moving all your data toward level 10 on Wolfram’s computable data scale.

The central idea for achieving the higher levels is to build a symbolic representational layer on the meaning and relationships of the data. That doesn’t require an upheaval in the data capture and data storage infrastructure but is about adding a layer that ensures that when you retrieve a value from a your data, you know what it means, how it relates to other values and what models, calculations or visualizations can consume it—and in a fully automated way.

Take a simple example: if you extract a 2 and a 3 from a database, can you do the operation “2 + 3”? If so, what does it mean? Well, if they represent inches and meters, we could, but the answer would not be 5. If they represent product IDs, the operation probably isn’t valid. But perhaps if they were IDs of investment portfolios, adding them together might be chosen to represent the combined portfolio. Doing this systematically so that high-fidelity digital twins or predictive models can consume data is what unlocks the open-ended, ad-hoc queries that an LLM could request.

In most organizations, that knowledge is patched with humans—librarians, business intelligence (BI) teams, analysts and others with similar roles. Not only is that expensive, it is also slow and the reason why most organizations only have near-real-time access to mission-critical data. And data deemed “less critical”? It will likely wind up languishing in a queue waiting for analysts’ attention.

Use Wolfram to Connect Your Dots

Smart business decisions come from making connections between disparate datasets. Take a retail company looking to streamline its supply chain: they’re not just looking at sales numbers. They’re diving into customer feedback, inventory levels and market trends. This holistic view uncovers patterns and forecasts demand with increased precision. And LLMs have the potential to crunch mountains of data to find insights your people could miss. But here’s the kicker: the advantages of LLMs can easily be limited by bad or messy data. If you feed them curated, high-quality data, they’ll give you recommendations that are spot-on. But if not? Bad analysis is worse than no analysis at all.

Your solution is Wolfram technology and our data curation team, which has a decade of experience in creating computable representations of enterprise data. We’re ready to help you on the journey toward enterprise AI.

Contact Wolfram Consulting Group to learn more about using Wolfram’s tech stack and LLM tools to generate actionable business intelligence.

Navigating today’s volatile business landscape without top-tier data is like trying to predict a hurricane with last month’s weather report. It’s not just reckless; it’s downright dangerous. Quality, up-to-date information is the Doppler radar for your business, helping you see through the unpredictable market conditions to make decisions that aren’t just reactive guesses but proactive strategies. After all, facts are as unyielding as the laws of nature: they don’t bend to our wishes or fears.

Cautionary tales underscore the critical importance of data accuracy in our interconnected, information-driven world. The 1999 NASA Mars Climate Orbiter failure serves as a stark reminder of how a simple unit conversion error, undetected due to inadequate data validation, led to the spacecraft’s destruction. Similarly, the 2008 financial crisis highlighted the dangers of relying on flawed historical data models that failed to account for the risks of subprime mortgages, resulting in a global economic downturn. More recently, the 2020 UK A-Level grading algorithm controversy demonstrated the pitfalls of not thoroughly validating the impact of data models on diverse student groups, leading to widespread perceived unfairness and policy reversal.

So what’s the solution? It’s not just one thing. Instead, multiple factors come into play: data breadth and curation, as well as its integration into a traditional programming environment or high-level AI/LLM analysis.

Enter the Wolfram Knowledgebase, a comprehensive data system that underpins Wolfram Language, with vast, curated datasets that are immediately computable—no data cleanup or excessive tagging/categorization is required because that’s already been done for you.

Data Scope

The Wolfram Knowledgebase is a comprehensive repository of curated data across multiple areas: from physical sciences (elements and molecules), medical data (gene properties) and astronomical information to economic indicators (stock prices), country statistics and cultural data, including historical events and notable personalities, along with internet-based metrics.

As Alan Joyce, Director of Content Development at Wolfram, describes it, “Basically, the Wolfram Knowledgebase covers pretty much every type of known information about the world that you can think of.”

This data is then organized into hundreds of different “entities” that represent either physical “things,” including types of animals (dogs and cats), movies (Casablanca to Barbie) and countries (the US and France), or mathematical and other scientific concepts such as space curves, mathematical functions and periodic tilings.

Plus, you’re not limited to only the data in the Wolfram Knowledgebase. You can also add entities for your own datasets and import that information to use with what’s already in the Wolfram entity framework.

Data Management

The breadth of available datasets is the first thing that sets the Wolfram Knowledgebase apart from other online databases. But this information is not just a hodgepodge collection of information from different sources. Instead, as detailed by CEO Stephen Wolfram, it’s carefully evaluated, selected and processed for integration within the Wolfram tech stack.

The process of transforming raw data into coherent computable data begins with collecting it from records, surveys and experiments, followed by converting it into digital formats such as .txt, .pdf and .xls files. The data is then systematically stored in cloud repositories with metadata and organized in structured formats using JSON and XML. Quantitative elements like quantities, dates and geolocations are represented in canonical symbolic form, and standard entities such as countries, species and chemicals are also given canonical symbolic representations. All entities, including custom ones, are uniformly consistent through normalization and manual or automated analysis. Computations for derived properties like interpolations, formulas and models are added, along with natural language mappings and access via tools like Wolfram|Alpha. The final goal is to ensure the data is suitable for repeated, systematic computations in Wolfram Language.

In contrast, a resource such as the Google-curated Dataset Search is more indiscriminate about indexing available online datasets, which can have limited (or no) manual curation, unlike the Wolfram Knowledgebase. And other information sources such as Quandl and OpenWeatherMap are more narrowly focused, which can make cross-referencing multiple dataset points more labor intensive due to the need for data alignment across multiple file types and categories.

Programming Integration

The final piece is the Wolfram Knowledgebase’s integration within Wolfram Language with sophisticated semantic representations such as Quantity and DateObject. This makes data analysis and visualization part of a unified framework as opposed to juggling data libraries from varied sources, multiple pieces of software and different programming languages.

“The Wolfram Knowledgebase isn’t just a big database that you can go and pluck data out of to put somewhere else,” says Joyce. “Its power comes from the fact that it’s all integrated with the Wolfram tech stack for analysis, visualization and deployment to the web.”

For example, with just a few lines of Wolfram Language code, you can easily generate an up-to-date UK air temperature map.

But the possibilities of data analysis and visualization with Wolfram Language extend even further. As George Danner, president of Business Laboratory, says about the software, “It can work on very large, very diverse datasets, including unstructured datasets, [and] can do just about whatever we ask and perform any kind of function. We’ve used statistical functions. We’ve used the optimization functions… to put together very fine-grained visualizations of the systems at hand, and these visualizations are critical for allowing our complicated models to communicate in non-mathematical ways to our audience, which are often company executives.”

Sustaining Success through Data-Driven Insights

Utilizing the Wolfram technology stack for data analysis can drive significant advancements and innovative solutions across multiple fields. For example, perhaps you need to cross-reference geographic, seismic exploration and imaging data to find new oil reservoirs. Or if you’re in power generation, maybe you want to use thousands of different inputs—everything from weather forecasts to plant repair time—to create predictive energy load models for power plants. And in finance, you may have to pull together data such as asset prices, cost of goods and inflation to analyze and predict economic trends.

One fact remains clear: in today’s digital age, quality data is the currency of success, and that requires multiple information streams to stay ahead of the curve. As you navigate the business landscape, remember that it’s not about weathering the storm—it’s about harnessing the power of data to create your own blue sky opportunities.

Contact the Wolfram Consulting Group to learn more about using the Wolfram tech stack, curated data and custom AI tools to generate actionable business insights.

It’s no secret: quantum computing has been poised to be “the next big thing” for years. But recent developments in the quantum ecosystem, including major investments by companies such as IBM, Google, Microsoft and others, are the best indicators that now is the time to begin preparing for potentially viable quantum applications—and to identify where and when to most effectively use them.

“Classical” computers operate in a binary fashion, processing information as either zeros or ones. But quantum computers? They leverage quantum mechanics so data can exist simultaneously in multiple states—zeros, ones or both. The result is a supercomputer that can explore countless possibilities at once and produce results in minutes or seconds instead of hours, weeks or years.

Projected use-case areas span multiple industries and market sectors. In finance, for example, quantum computing could be utilized for portfolio optimization, risk management and algorithmic trading. In healthcare, it could enable advancements in drug discovery, personalized medicine and disease modeling. In logistics, it could optimize supply chain management, route optimization and scheduling.

At the same time, there’s a gap between what quantum can do at the current moment versus anticipated capabilities in the next three to five years.

“In each real-world application, the first step in solving the problem is translating it to a computational problem. After that’s done, you need to choose—or develop—an algorithm for solving that problem. And because these applications are things people do every day, there are clearly a number of ‘classical’ computing algorithms that are known—and still being explored—for solving the associated problems,” says John McNally, Wolfram Academic Innovation Solutions Developer. “The catch is that the scale of certain real applications can become too great for even the best classical computing resources to successfully carry out any known classical algorithm to reach a full solution.”

At this point, quantum algorithms and the hardware needed to run them become interesting. And there are important questions to address, in particular how to scale up quantum systems following practical fault-tolerant approaches.

And this is where the Wolfram Quantum Framework comes in.

Wolfram Quantum Framework

On its own, the Wolfram Quantum Framework is not a quantum computer. Instead, it is a set of tools to model quantum circuits and design algorithms. Then, after you’ve analyzed a quantum system using classical means, the Framework also gives you tools to automatically translate your models into representations that can be run on quantum hardware through service connections with Amazon Braket and other providers:

• Classiq
• Infleqtion
• Q-CTRL
• QuEra
• Strangeworks

The Wolfram Quantum Framework’s key advantage, however, is its seamless integration with Wolfram Language. This includes optimized numerics and symbolics that offer a streamlined approach to quantum computation.

“Aside from working numerically, you can specify noise channels, gates or even elements of a Hamiltonian as symbolic parameters,” McNally says. “This allows you to get exact formulas out of your analysis rather than numerical simulations only. Plus, it seamlessly integrates with other quantum platforms, even across different programming languages.”

Wolfram’s symbolic computation offers distinct benefits over numerically based programming languages by enabling the calculation of “exact” solutions to complex problems. For academia, this means a deeper understanding of quantum principles and the ability to explore new frontiers in research. But for businesses, prospective bottom-line returns include potential cost savings through more efficient algorithms, improved accuracy in simulations and predictions, and faster innovation cycles.

The Wolfram Quantum Framework continues to evolve with new features to expand the Framework’s support for different computational models, such as tensor networks and stabilizer formalism, as well as to accommodate a wider range of quantum tasks. Additionally, plans are underway to introduce enhancements to improve usability, scalability and compatibility with emerging quantum hardware technologies.

The Future of Quantum

Navigating quantum is more than thinking about quantum algorithms alone: it also means understanding the world of classical algorithms to identify when practical problems make contact with the rapidly developing world of quantum hardware. This uniquely positions Wolfram Research to address challenging quantum problems and develop quantum utilities, particularly considering its longstanding development of classical algorithms and well-developed footprint in academia.

In a larger sense, the development of quantum capabilities is going to result in a great divide: the haves and the have-nots, or more importantly, the dids and the did-nots. While quantum hardware is not yet where it’s going to be for use on an industrial scale, the time is now to begin planning for its use when it is. And while this may seem a bit nebulous, vision is the art of seeing what is invisible to others, and those who understand how quantum will benefit them stand to reap the most significant rewards.

Contact the Wolfram Consulting Group to learn how the Wolfram Quantum Framework can provide insights and tools for innovation.