Artificial General Intelligence (AGI) might have an efficiency problem

Published: Sunday, December 7, 2025 at 12:33 pm

AI Deployment: Navigating the Path to Practical Value

Ruchir Puri, chief scientist at IBM, sat down with The Deep View to discuss the problems with the existing perceptions about Artificial General Intelligence (AGI).

Puri had this to say (and much more):

..."If you focus on artificial general intelligence, it drives the progress in technology. It makes the underlying technology better and better and better. The misguided part is that I don't need artificial general intelligence to write an email.

I'll give you an example. Let's contrast artificial general intelligence with real intelligence. Real intelligence, which is our brain, lives in a 1200-centimeter cube, consumes 20 watts and runs on sandwiches. That's the efficiency of a human brain. A single GPU board of Nvidia Blackwell chips consumes 1200 watts, 60 times more, and you need tens of them, if not hundreds of them. You're talking about a difference of three orders of magnitude in efficiency.

What I'm saying is: don't use artificial general intelligence to solve very specific enterprise tasks. Usefulness implies solving a problem with the cost that I need, with the efficiency that I need and where I need it."...


Businesses seeking to leverage artificial intelligence are facing a critical challenge: extracting tangible value from their AI deployments. Experts are highlighting key strategies to navigate this complex landscape and avoid common pitfalls.

A central theme emerging from discussions is the importance of focusing on "artificial useful intelligence" – AI that delivers practical results. The initial step, according to experts, is selecting the right task. Many AI projects fail because the chosen problem is either too ambitious or not well-defined. Careful consideration of the specific task is crucial for success.

Following task selection, the establishment of clear evaluation metrics is paramount. Without well-defined success criteria, it becomes impossible to gauge the effectiveness of an AI implementation. Businesses must clearly define what constitutes a successful outcome before launching their AI initiatives.

Optimization is another critical element. Experts caution against expecting immediate miracles. Instead, they advocate for a phased approach, emphasizing the importance of learning and refining the implementation process over time.

A common mistake enterprises make is misaligned expectations. Companies often choose problems that are either too complex or too simple, leading to disappointment. Finding the "sweet spot" is essential for building trust in the technology. Another key factor is the availability of the right data. Enterprises should focus on domain-specific knowledge to guide and solve tasks effectively, rather than relying on general intelligence models. This approach is seen as crucial for achieving the desired return on investment. Furthermore, educating the workforce and building the necessary skills to integrate AI into existing workflows is considered essential for long-term success.

BNN's Perspective:

The insights shared underscore the need for a pragmatic approach to AI deployment. While the potential of AI is undeniable, businesses must prioritize practical applications and realistic expectations. Focusing on specific tasks, establishing clear evaluation criteria, and fostering a skilled workforce are essential for realizing the true value of AI. This measured approach, prioritizing utility over hype, is likely the most sustainable path forward for businesses seeking to integrate AI into their operations.

Keywords: AI deployment, artificial intelligence, artificial useful intelligence, business, task selection, evaluation, optimization, expectations, data, workforce education, ROI, IBM, open ecosystems, models, workflows

Full Story