Jumping Someone Else's Trained Model
by Jay CuthrellShare and discuss on LinkedIn or HN
This week’s musical inspiration in title and lyrics:
This week we take a look at a reader request for GenAI stories that are unrelated to platform engineering.
Getting Informed
Fudge Sunday has been on quite a run and the past 50 issues represents just under a year of time since joining IBM — time flies. 📊
I am incredibly thankful for the readers, their comments, their suggestions, and _their questions_. 🙏
One reader of Fudge Sunday from the analyst community posed some interesting generative AI (GenAI) questions in response to #209 — AI Feel You.
Who sees GenAI benefits outside of technologists?
What practical GenAI benefit can be experienced and grasped by business leaders?
When will everyone begin to benefit from GenAI?
Before I try to answer these questions, please consider these selections for reading 📖 and watching 📺:
📖 AI Will Change Our Current Organizational Structure in which Sandi Bensen looks at how individuals and companies can anticipate the changes ahead.
📺 Generative AI for business in which Dario Gil demystifies an open data lakehouse architecture — and the bring your own data.
It won’t take you long to learn the new smile 🎶
First, I don’t think that GenAI has reached a fever pitch phase of invite codes and developers scrambling for access to tools. In fact, in the past two weeks I received invitations to different GenAI options for testing that are not specific to software development.
To be clear, there are two very specific non-software development oriented stories to tell about. Each story relates to answering questions based on something most companies have — content.
Two weeks ago, my friends at SiliconANGLE & theCUBE bumped me to the front of the waitlist for access to a private beta of their home grown generative AI model which can generate responses based on their unique analyst coverage over the years.
Takeaways:
- Clever “Related Clips” to see what is likely to be useful for embedding as contextually relevant further viewing
- I can’t wait to see how this approach impacts thinking on “creator oriented search” against the body of work as compared to the default search options for YouTube (which are typically pretty awful)
Then this past week, HFS invited me to kick the tires on a trained generative AI model powered by Humata.ai with access to a limited number of banking, financial services, and insurance ( BFSI) industry publications.
Takeaways:
- Clever “Highlighting” of text that aligns to the response along with hyperlinked footnotes with copy / feedback UI
- Suggestions for what to look at next are probably a way to upsell licensed access to materials today — but imagine these being a preview for what could become a vasty improved version of support knowledge base search on most company websites
If you abstract these two examples to simply the content generated by a company, you can see likely GenAI benefits for anyone seeking an answer from a company (that offers potential for a notable service improvement that could displace / replace brute force website keyword search of the past two decades).
There’s even a $30 per user per month price established for access to the technology to unlock and unleash the capabilities of GenAI.
https://www.techmeme.com/230718/p17#a230718p17
Back to the questions…
When will everyone begin to benefit from GenAI?
Expect the price per month to drop as access becomes democratized, commodified, and subsidized. So, it’s probably worth wondering where and how the GenAI related prompts will direct end users to places where answers are more easily discovered.
You’ll have to adapt or you’ll be out of style 🎶
Second, editors are important. We’re still in the rainbow table exhaustion automation phase of GenAI.
Back to the questions…
Who sees GenAI benefits outside of technologists?
At the risk of saying I’m not a technologist, I do.
What practical GenAI benefit can be experienced and grasped by business leaders?
Let me explain as a non-technologist… I’m busy. I work in a business.
For this issue of Fudge Sunday, I used the “Generate” button to create a summary from Buttondown that makes use of a very good business use case for GenAI… a clear, crisp, concise summary of what follows.
That’s not too bad. Then again, this is only a 500-1000 word newsletter issue and I did take editorial controls before just letting that go as-is.
Editors and editing are still non-negotiable requirements for GenAI.
But are there are other indications that more of these “Generate” buttons are coming to a familar application / web / mobile interface in the near future?
Yes. Count on it.
Where will we see it happening?
I’d look to companies like Glitch, CodeSandBox, and Developer Sandbox as harbingers of things to come. Remember, our low-code and no-code future is approaching.
There’s an old saying… If you can’t beat them… partner with them in a press release and run as fast as you can.
https://www.techmeme.com/230718/p28#a230718p28
So, what will be the next big thing in GenAI stories?
Until then… Place your bets!
Disclosure
I am linking to my disclosure.
🤓
Topics:
✍️ 🤓 Edit on Github 🐙 ✍️
Share and discuss on LinkedIn or HN
-
Get Fudge Sunday each week