Banyan - September 2023

AI Can Predict the Weather, What About Financial Forecasts?

Artificial intelligence (AI) transforms how work is organized, allowing organizations to process insights at scale. 

Its far-flung applications and everything, everywhere, all-at-once capabilities, that when compared to traditional methods offer huge advancements, have generated enterprise excitement across industries as well as stimulated the broader marketplace. 

And now, the technology can even successfully predict the weather. 

A new AI model from Google DeepMind — GraphCast — was able to predict weather conditions around the world up to 10 days in the future, outperforming the current gold-standard meteorological systems across nearly all (90%) of the 1,380 key metrics measured.

“GraphCast is a key advance in accurate and efficient weather forecasting, and helps realize the promise of machine learning for modeling complex dynamical systems,” wrote the DeepMind team. 

The weather prediction AI model didn’t just beat traditional methods on accuracy alone: The AI was also able to surface results much faster and for a fraction of the compute cost and energy consumption. 

It was trained on 40 years of historical weather data and can produce a 10-day forecast in just one minute. 

And the model’s resounding success in meteorology poses the question, Just what else can AI forecast better than today’s existing models?

Read also: Tailoring AI Solutions by Industry Key to Scalability

Supercharging Financial Forecasts and Fraud Detection

PYMNTS has previously reported on whether AI tools might be the perfect fit for a CFO office tasked with safeguarding the financial health of the organization while simultaneously optimizing efficiency and return on investment.

After all, advanced AI technologies are increasingly playing a role in streamlining payments, bringing automation and efficiency to an industry traditionally reliant on manual processes.

Examples of where AI technology can have an immediate impact include automating billing and accounting reconciliations, updating customer relationship management (CRM) and enterprise resource planning (ERP) systems in real time without the need for manual intervention, and extracting information from legal and contractual documents using natural language processing (NLP).

Increasingly, firms are turning to the innovation for help with real-time fraud detection and to support financial forecasting and cash flow management workflows. 

“In [today’s] higher-rate environment where the cost of capital has increased, every penny is valuable. And that includes having good visibility and managing your working capital to the highest degree of precision possible,” Pat Dillon, CFO at Flock Freight, told PYMNTS. 

Amias Gerety, partner at QED Investors, told PYMNTS last spring that he believes, at least right now, the most interesting areas where AI is being applied lie in fraudrisk management and underwriting because those are industries where practitioners are sophisticated and can use AI as a tool to get fast answers while ensuring that results are accurate.

Not knowing whether a transaction was fraud or a false positive cost U.S. eCommerce merchants $81 billion in 2023 according to PYMNTS Intelligence. By using transformer neural networks and multi-head attention mechanisms to collate and produce relevant information sets in real time from vast data, AI solutions can supercharge areas like fraud detection to a degree impossible for humans to tackle alone. 

Read alsoWalled Garden LLMs Build Enterprise Trust in AI

Safeguarding Security-Critical Operations

Two key details of the AI weather model were that it outperformed on 90% of measured metrics, not 100%, and that it was purpose-built on decades of hyper specific, contextually narrow training data. 

As PYMNTS has reported, at the center of many business use concerns around the integration of generative AI solutions lie ongoing questions around the integrity of data and information fed to the AI models, as well as the provenance and security of those data inputs.

For example, PYMNTS Intelligence finds that 72% of lawyers doubt the legal industry is ready for AI, while just 1 in 5 believe in the advantages of using AI surpass the disadvantages.

These concerns are only heightened when it comes to critical areas like finance and payments. 

“We’ve got to be careful how we use this technology in a compliant manner,” i2c CEO and Chairman Amir Wain said to PYMNTS, cautioning against rushing full speed and embracing the technology. “We cannot be at the bleeding edge of technology dealing with money and funds. … We need to put a compliant framework around the tool.”

AI and machine learning (ML) have the potential to provide insights that aid decision-making and can assist businesses in identifying payment trends, anomalies and potential issues, allowing for proactive measures to be taken, Joe Pergola, CFO at Connatix, told PYMNTS.

But to do so effectively and at scale, enterprise AI solutions need to be trained on the right data and applied in a controlled and compliant context. Simply plugging into an API from a buzzy AI startup won’t cut it for a multinational organization, or its finance and accounting teams. 

That’s why companies like JPMorgan are seeking regulator guidance as they build out new AI products that touch on safety-critical areas, and others like Visa are launching AI Advisory Practices. 

“Enterprise use of AI has to be accurate and relevant — and it has to be goal oriented. Consumers can have fun with AI, but in a business chat or within an enterprise workflow, the numbers have to be exact, and the answer has to be right,” Beerud Sheth, CEO at conversational AI platform Gupshup, told PYMNTS. 

And as Shaunt Sarkissian, founder and CEO of AI-ID, told PYMNTS in May: “No matter the ways and means in which AI is being harnessed, it’s incumbent on firms to mull how they can enhance value rather than just chase a trend.”