According to the founder and CEO of Technology Company Pegasystems, technology leaders are baffled by the hype and the “dissonance” of confusion over the benefits of artificial intelligence (AI).
Known for his chess and ping-pong prowess, Alan Treffler runs a $1.5 billion sales technology company and spends much of his time meeting clients, CIOs and business leaders.
“I think CIOs have a hard time understanding all of the buzzwords, hype and confusion that exist,” he said.
“The words AI and agents are thrown into this great dissonance. They don't know what it means. I'm constantly listening to it.”
CIOs are under pressure from CEOs and we are confident that AI will provide something of value.
“CIOs are really hungry for practical and practical solutions. If they don't have them, many of them do a lot of experimentation,” Trefler said.
Companies are looking at large language models to summarise documents, stimulate knowledge worker ideas, and generate the first draft of reports. All of these save time and make people more productive.
The risks of Rouge AI
However, Trefler said companies are wary of loosening critical business applications into AI.
“There's a lot of fear about passing things over to something that no one knows exactly how it works. That's an absolute play situation when it comes to the general AI model,” he said.
Trefler is struggling with large tech companies driving AI agents and large language models for Business-Critical applications. “I think they went on a convenient but shortsighted path,” he said.
“If these operations must be predictable, reliable, accurate and fair to the client, the idea of you handing over important business operations to agents is not just short-term, but structurally problematic.”
One of the problems is that the generated AI model is very sensitive to the construction of the data being trained and the prompts used for instructions. Small changes to the prompt or training data can lead to very different outcomes.
For example, a business banking application may know that the customer is a little more wealthy or a little poorer than expected.
“It's easy to imagine a quick decision to change the rates charged, whether that was what the institution wanted or whether it was legal in accordance with the various regulations that lenders have to follow,” Trefler said.
Thinking ahead is better than thinking with your feet
According to Trefler, Pega has taken a different approach to adding AI to other technology suppliers to its business applications.
Rather than using AI agents to solve problems in real time, AI agents think ahead of time.
Business professionals can use them to co-design business processes to do anything from assessing loan applications, providing offers to valued customers, or submitting invoices.
Companies can continue to deploy AI chatbots and bots that can answer queries over the phone. Their job is not to resolve solutions from scratch for every inquiry, but to determine which is the appropriate pre-written process.
As Trefler stated, design agents can create “dozens and dozens” workflows to handle all the actions a company needs to take care of its customers.
“Semantics allows you to use natural language models to handle the miracle of getting the language right, but there are ways to implement reliable, predictable regulatory approval methods to connect that language to your workflow,” he said.
AI suitable for work
Large-scale language models (LLM) are not necessarily the right solution. Trefler demonstrated whether ChatGPT 4.0 attempted to solve the chess puzzle. LLM repeatedly suggested impossible or illegal moves despite Trefler's modifications. Meanwhile, another AI tool, Stockfish, which is a dedicated chess engine, solved the issue immediately.
Another drawback of LLMS is that it consumes a huge amount of energy. So, if AI agents are inferring during “running time,” Trefler said, they will consume hundreds of times more power than AI agents simply choosing from a pre-determined workflow.
“ChatGpt is inherently very consuming… when you're answering your question, that firing is literally a few hundred million to trillion nodes,” he said. “I need all of that [large quantities of] electricity. “
Using employee payroll bills as an example, Trefler said a better alternative would be to generate 30 alternative workflows to cover the key variations found in payroll bills, for example.
It gives you “real idiosyncraticity and real efficiency,” he said. “And that's a very different approach to handing over the process to the machine using the prompt and having the machine reason about it every time.”
“If you've come down to the philosophy of using graphics processing units [GPU] Get 200 for workflow engine to run workflows to create workflow enginesth It's electric because there's no reasoning,” Treffler said.
He clearly sees the increasing use of AI will have a major impact on the job market, and the entire employment category will disappear.
For example, the need for translators could be exhausted by 2027 as AI systems become better with the translation of spoken and written languages. Google's real-time translators have already improved, saying “terrifyingly good.”
Pega currently plans to work closer together with a network of system integrators, such as Accenture and Cognizant, to provide AI services to businesses.
Major expansions
The initiative launched last week allows system integrators to incorporate their own best practices and tools into PEGA's rapid workflow development tools. This move means Pega's technology will reach a wider range of businesses.
With programs equipped with Pega Blueprint, System Integrators can deploy customized versions of BluePrint.
This tool can be used to reverse aging aging applications and replace them with the latest AI workflows that can be run on PEGA's cloud-based platform.
“The idea is that we are trying to make this Blueprint Agent's design approach available not only through us, but through many key partners that supplement our intellectual property,” Trefler said.
This represents a major expansion of PEGA. It mainly focuses on supplying technology to hundreds of clients and represents the Top Fortune 500 companies.
“We've never done anything like this before, and I think it will lead to a huge change in how this technology can be put into the market,” he added.
