40% Smarter with GenAI

As business people, when we speak of GenAI, we always think of the “possibilities” - efficiency, productivity, automation, cost savings, and so on!

Egged on by the enthusiastic optimism of tech entrepreneurs (including yours truly), we think of the possibilities as “unprecedented,” “best yet,” etc.

Is GenAI worth the hype?

The Economist doesn’t think so.

To make the point about the reality of GenAI’s impact, the magazine compares it to the rise of computers in the 1960s. It says, “In 1987 Robert Solow, a Nobel laureate, famously quipped that ‘you can see the computer age everywhere but in the productivity statistics.’”

Haha. Thankfully, the promised “transformation” came in the 1990s.

Perhaps, similarly, the impact of GenAI is too early to measure. What, then, can we learn from its parallels with digitization?

It’s here if you’d like to find out.

Data processing to go from a SQL-centric workload to an AI-centric one

If you’re like any of the enterprises we talk to, you’re using SQL queries to build reports/dashboards to get insights from your data. And you’ve hit a wall.

As Robert Nishihara of Ray and Anyscale points out, “SQL queries aren’t going to operate on PDFs, videos, images or arbitrary text.” He argues that, for example, in a sales organization, all of the insight is in meeting notes and sales calls - i.e., unstructured - that are currently unusable without AI.

He imagines that in the future people will use AI to “read the data, reason about it, and draw conclusions.” More of his ideas around data for AI are in this podcast here.

At some level, the future he imagines is already here, though.

Generating custom proposals based on past data

One of UAE’s largest event management companies had one critical process with over 40% repeatable elements: Proposal creation.

They came to Tune AI to explore possibilities around automating this based on 20 years of their own custom proposals data - most of which were unstructured in Word documents and PDFs.

Here’s the story of how we increased the scale and efficiency of creating customized proposals for large, high-value events without adding headcount.

Read Tune AI case study

The more you GenAI, the more you need human judgment

Before I sign off, I want to leave you with serious food for thought about the autonomy of GenAI. Last week, I wrote about how tech leaders seem to think artificial general intelligence (AGI) is just a year away. This means that, at some level, we all hope that GenAI will do things independently.

However, as we stand today (and for the foreseeable future), “generative AI is still just a prediction machine,” write researchers in the latest edition of HBR.

Even for the most accurate predictions, companies need human judgment to take the right action. “Organizations with good judgment will thrive as AI diffuses,” they say.

Ironically, good AI needs good human instinct. More at the HBR website here.

With that, I say bye for this week.

Until next time, stay tuned.

Best,
Anshuman Pandey

P.S. Llama 3.2 Vision Language model is now available on Tune Chat