Generative artificial intelligence (AI) has most companies looking to buy and integrate, not build, solutions.
That’s because the sheer computing cost of running large language models (LLMs), much less the considerable expertise required to build them, leaves most enterprises with little say in the matter.
Complicating the situation for firms looking to capture an innovative edge is that at the center of many business use concerns around the integration of innovative 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.
Popular incumbent AI solutions like OpenAI’s ChatGPT are operated through online data centers that can expose sensitive information.
Taiwan’s Asustek Computer Inc. (Asus) is launching a new subscription-based AI service called AFS Appliance that will let companies keep control over their data.
Read also: Microsoft Fends off AI Data Concerns With Private-Server ChatGPT Solution
The Asus AI subscription, which will reportedly cost between $6,000 and $10,000 per month, will run on a generative AI supercomputer cluster using Nvidia graphics processing units (GPUs), and all of the hardware will be installed on site at client offices, giving the subscriber control over data security.
“Agile companies will take advantage of AI and boost their position,” said Nvidia CEO Jensen Huang. “Companies less so will perish.”
Nvidia has seen its stock price soar alongside the hockey stick demand for its products, which include computer chips used for building and training AI models.
The company now has a market cap of more than $1 trillion, primarily based on its role as a critical component of tomorrow’s AI-driven operating environment.
Research in the report “Bridging the Gap Between Subscription Conversion and Retention,” a PYMNTS and sticky.io collaboration, found that subscription cost drives 56% of all cancellations. To combat this, 97% of top-performing subscription providers offer the ability to pause or modify their subscriptions.
It remains to be seen whether the Asus AI subscription, which will run on a bespoke LLM the company says is equivalent to OpenAI’s ChatGPT 3.5, will buck this trend.
See also: Is It Real, or Is It AI?
The focus for many businesses hoping to capture the next-generation operational efficiencies native to AI integrations across historically manual processes is centered around establishing the appropriate technical infrastructure to be able to embed AI applications into back-end and end-user-facing systems.
Still, headlines being made by a lawyer who used generative AI to file his court briefings show that the technology by itself is no silver bullet, and it will require certain industry-specific guardrails to be effective.
“AI is here to augment people … developing an AI algorithm itself is not that hard,” Erik Duhaime, co-founder and CEO of data annotation provider Centaur Labs, told PYMNTS. “A company with a great AI solution needs to have great data and a great data annotation process, and that boils down to having great people.”
In the lawyer’s case, the AI he turned to invented made-up decisions titled Martinez v. Delta Air Lines, Zicherman v. Korean Air Lines, and Varghese v. China Southern Airlines to support his court briefing — all of which were determined to be pure fabrications by the presiding judge, Kevin Castel of the Southern District of New York.
Read also: AI Regulations Need to Target Data Provenance and Protect Privacy
Deepfakes are widely considered to be the biggest AI-related threat, while AI technology’s propensity for “hallucinations,” or the creation of original — and false — content, remains the biggest limiting factor bottlenecking widespread acceptance and integration of generative AI tools.
“There is a lot of value [around generative AI capabilities], but the key question is when can we use it without the fear of bias, and where this information is coming from?” Bank of America CEO Brian Moynihan said in April. “We need to understand how the AI-driven decisions are made…”
While generative AI can place expert knowledge into the same room with every single person who needs it, particularly for institutions like banks or hospitals, the importance of getting the information right when it comes to applying AI is mission critical.
At a high level, generative AI has the potential to create a new data layer, like when HTTP was created and gave rise to the internet beginning in the 1990s, Shaunt Sarkissian, founder and CEO of AI-ID, told PYMNTS. As with any new data layer or protocol, governance, rules and standards must apply.