Reflections & analysis about innovation, technology, startups, investing, healthcare, and more .... with a focus on Minnesota, Land of 10,000 Lakes. Blogging continuously since 2005.

Category: AI (Page 2 of 3)

#EnterpriseAI: General Mills Is ‘Transforming With the Power of Cloud and AI’

I attended an event yesterday at the headquarters of General Mills to hear how the company has partnered with Google Cloud over the past few years to transform itself in the age of cloud and AI. No doubt, hearing how any Fortune 500 could pull off something like this would be quite a story, I figured, let alone a company that’s been around for 155 years(!). Yes, we’ve been known to build ‘em to last here in Minnesota.

I pulled into Betty Crocker Drive in Golden Valley MN about 8:15 am and drove into the sprawling, grassy campus, with several gleaming buildings. I hear it’s not all that busy these days, however, what with remote work. But it may be the most impressive facility of any of our giant local companies for holding a big event like this. The crowd looked like it numbered 300 or more. Being that I’m a big fan of in-person events (but somewhat skeptical of AI hype), I had to check it out. Plus I felt an obligation to be there in light of my role as a board member of MinneAnalytics. (We love data and AI!)

So, what I thought I’d do is share my notes with you, along with some photos, plus stand-out quotes from the executive panel that followed the morning speakers.

Notes from the Opening Keynote by Rich Rubenstein, VP-Data & Analytics

Rich Rubenstein

Rich Rubenstein

– All General Mills data is now on the Google Cloud Platform today.
– The company uses Vertex AI.
– “This is not a science-fair project. We use AI every single day.”
– The focus is on innovation — “in products, in go-to-market, and in how we connect with consumers.”
– On the topic of AI hype: “There’s a tremendous amount of light and heat in this space. But we know we have a solid foundation. We’ve invested in someone [Google] that we know can take us a long way.”

 

Some Remarks from Jenny Hon, Senior Director, Strategy & Enterprise Architecture

– The cloud transformation journey is based on the company’s purpose: “To be the undisputed leader in food.”
– “All signs indicate that our ‘Accelerate Strategy’ is working.” accelerate strategy
– How we win is by doing four things: “Boldly building brands, relentlessly innovating, standing for good, and unleashing our scale.”
– She noted that General Mills has nine billion-dollar brands.
– The company used Cloudera for a long time, “but it wasn’t enough to take us where we wanted to go.”
– What is cloud for General Mills? They focus primarily on infrastructure as a service (IaaS).
– Developing a FinOps practice has been a major focus — “to ensure business unit cloud spends don’t blow their budgets.”

Some Remarks from Josh Moe, Senior Manager, Enterprise Architecture

– “Our aspiration is to lead CPG in data and analytics.”
– The company began in 2019 with a big bet in data analytics with Google Cloud.
– They had looked at two major vendors [the other was Microsoft Azure].
– “We moved SAP from on-prem to cloud. There was a lot of complexity in that.”
– “We did a major transformation and data migration in 24 months. Even Google doubted we could do that.”
– Specifically, the data lake migration took 15 months, and the data center migration took 18 months.
– “We did a full rebuild of how we do SAP.”
– The company’s “Analytics Enterprise Data Warehouse” is on BigQuery. “It’s the enabler of the work we’re doing in AI.”
– “We’re very intentional about our data. Do we own it? Or just subscribe to it?”
– The company has “Automated Self-Service Infrastructure Processes” that provide a simple and secure experience for its data scientists, product teams, and developers.

Some Remarks from Hanna Gordon, Director, Digital & Technology Data Science

– She started with General Mills in 2016, and has managed rapid growth of its AI/ML team.
– In 2019, the team began work with Google Cloud and Google AI.
– Forbes has called her AI team one of the fastest-growing in the country.
– “We have 6 million models in production today.” [Are you kidding me!?]
– Their first use case went live in 2020.
– In 2022, the company’s migration was complete and they began building a machine learning engineering team.
– In 2023, her team has been optimizing workflows, building MLops, and working with generative AI, “like everyone else.”
– The team now numbers 75 in the U.S. and India.
– They now have three core roles: data scientist, ML engineer, and MLops analyst (the most recent role).
– She showed a slide (see “Model Growth”) that listed their number of models per year, and the number of predictions per month — now 500 million(!) from the current 6 million models.

Following the above talks, after a break, there was an executive panel moderated by Saher Asad, Director of Engineering for Google Cloud. The participants were:
– Jeff Young, Chief Data Officer, Prime Therapeutics
– Jason Staloch, VP-Digital Core, General Mills
– Mark Langanki, CTO, C1 (ConvergeOne)

A few highlights:
– Staloch: “Gen AI is the third big thing [transformational technology], after the Internet and the iPhone, that really started on the consumer side.”
– Young: “The first thing we look at with gen AI: is it safe and secure?… We have to be able to wall off our data.”
– Staloch: “There’s a closer relationship now between business and IT. We can tell the story better now.”
– Asad: “Gen AI has taken a lot of the complexity out” [of the AI journey].
– Langanki: “One use case we found, when a conversational chat answer is ‘I don’t know’, we feed that chat into gen AI to find out what the real question was.”
– Young: “Building internal literacy is important for us.”
– Staloch: “The focus should not be on finding a use case but, rather, what is the problem?”
– Langanki: “The conversation on a bot must stay private.”
– Asad: “There’s a big rush now in gen AI, but it’s not the answer to every problem.”
– Staloch: “We’re introducing a private LLM for employees.”
– Young: “The biggest challenge now is how to deploy AI at scale.” His advice: “Start slow. Make sure your data is as high quality as possible,”

One final point cited by Asad, the moderator: Gartner has found over time that only 60% of IT pilots make it into production. The firm also reported recently that it predicts 91% of future IT projects will involve AI. You do the math.

Thanks to General Mills and Google Cloud for putting on this event!

By the way, there were several more technical and hands-on sessions in the afternoon, but I wasn’t able to stay for those. I’ll be interested to hear from colleagues about those sessions.

Have you asked ChatGPT to write your resume or bio?

I got an inquiry from a reporter recently who wanted to know if myself or any of my colleagues could suggest good ChatGPT prompts to use to write a resume. (I sent her to one of my contacts who’s into career counseling, thinking he might weigh in.) I had nothing to offer her myself, but her request got me wondering… what would happen if I asked this all-knowing, all-seeing platform to write a bio for me. I don’t need no stinking resume, but bios — yeah, they’re cool. I’ve had to write them in the past in several lengths for various purposes.

AI-generated image of robot writing flowery language

Robot writing flowery language, created by me using OpenAI’s DALL-E.

So I did just that – submitted my request to ChatGPT. I figured, being that my online presence now spans some 28 years, including tons of writing over that time, 20 years on LinkedIn (as of this summer), and 16 years on Twitter (along with other sundry social-media sites I care less about), this surely should not be a big deal for the current King of AI to handle, right? It has total, complete access to that online presence of mine (at least through the fall of 2021).

Well, I found out quickly. It took only seconds to get results. I tried several approaches — asking for bio lengths of 300, 200, and 100 words. And I made it a point after my first tries to reference LinkedIn and Twitter in my prompt, thinking that surely would provide more grist for the mill, wouldn’t you think? (Well, the results didn’t change much that I could discern — which surprised me.)

Anyway, below is the first one I got — the 300-word version. (It’s actually 366, so apparently ChatGPT doesn’t know how to count – haha!) What got me was how over-the-top gushy it was! I would never be able to write one like this myself. It reminds me of the types of author bios that publishers put on book jackets… you know, to make the person sound like he/she walks on water? And, surprisingly, it contains no specific details or references about my background — just a bunch of flowery language:

Graeme Thickins is a renowned marketing strategist, prolific writer, and influential industry figure who has left an indelible mark on the world of marketing. With an extensive online history, a wealth of writings, and a stellar reputation, Graeme has established himself as a trusted authority and sought-after resource in the field.

Through his blog, articles, and social media presence, Graeme has consistently shared valuable insights, trends, and thought-provoking ideas, showcasing his deep knowledge and expertise in marketing strategy. His writings cover a wide range of topics, including brand management, digital marketing, integrated campaigns, and market research, among others. Graeme’s ability to distill complex concepts into accessible and engaging content has garnered a significant following and earned him a reputation as a thought leader in the marketing community.

Known for his strategic mindset and analytical approach, Graeme has a keen eye for identifying emerging market trends and consumer behaviors. He combines this astute understanding with his extensive experience to craft innovative marketing strategies that drive results. His knack for blending traditional and digital marketing channels has garnered attention, and his campaigns have consistently delivered impressive outcomes for the organizations he has worked with.

Beyond his writings and marketing expertise, Graeme’s reputation is also built on his collaborative nature and willingness to help others succeed. He actively engages with his audience, offering guidance, mentorship, and constructive feedback. His generosity in sharing his knowledge and insights has earned him the respect and admiration of colleagues and industry peers alike.

Graeme’s reputation extends to speaking engagements and industry events where he shares his expertise and experiences. His presentations are renowned for their practicality, delivering actionable takeaways that resonate with attendees. He is often sought after as a speaker, panelist, and workshop facilitator due to his ability to educate, inspire, and empower marketing professionals at all levels.

In summary, Graeme Thickins is a highly respected marketing strategist, accomplished writer, and influential industry figure. Through his online history, extensive writings, and sterling reputation, he has proven his ability to navigate the ever-evolving marketing landscape and consistently deliver impactful results. His dedication to knowledge sharing and fostering professional growth has made him a trusted resource and an invaluable asset to the marketing community.

But, damn, I sure sound like an impressive guy, huh? And it didn’t cost me a dime.

Ain’t AI great?  Well now, wait a minute — let’s hold off deciding that …

——–

p.s.  By the way, it’s interesting that HR people already have ways to defeat AI-written resumes. Humans win again!

“Expose ChatGPT Resumes and Uncover Real Talent Using These 5 Effective Strategies”
https://www.entrepreneur.com/leadership/expose-chatgpt-resumes-and-uncover-real-talent-using-these/448146

 

My 2023 Predictions Post

I asked AI to help me peer into the future. As is my custom, that resulted in me doing my top ten predictions:

robot contemplating the future

Robot contemplating the future, created by me using OpenAI’s DALL-E.

1) I predict, as things move so quickly these days, it will become much more common to see monthly predictions posts popping up. A year is now like a decade.

2) For the first year ever, there will be more articles written about AI than the number of orders of fries McDonalds will sell.

3) More than 60% of Americans will pay no income tax — oh, wait, that’s already the case.

4) Elon Musk will be richer at the end of the year than he is now. But he won’t care.

5) Twitter will go public (again), which will restart the IPO market. The media will refuse to give Elon Musk credit for that, but, you guessed it, he won’t care.

6) The NFL will end the year with more money than God — but God, not a huge football fan, will demand a recount.

7) The number of minutes of commercials on the average NFL game will be more than double the number of actual game minutes.

8) Gopher football will lose to Iowa again. (Kirk Ferentz beats PJ Fleck like a rented mule.)

9) I will not stop talking about football.

10) I might stop doing predictions posts.

p.s. Okay, wanna see what AI would really predict for 2023? Here’s a fun piece from a UK site:
We Made AI Predict What Will Happen In 2023, Here’s What It Said.

——

UPDATE 1/2/23:
Speaking of restarting the IPO market, I see Forbes just published a piece naming their picks for companies that are likely to go public in 2023. They don’t have the balls to pick Twitter, as I do — haha! But one of theirs has Elon’s name all over it: SpaceX. There you go — one of the reasons why I said Elon will be richer at the end of the year! I posted the Forbes piece in my new Flipboard magazine, “IPO COMEBACK.”

I Couldn’t Help Myself – I Did It Again

When you sign up for a one-year subscription to the Lensa App (yes, I got sucked into that), you are naturally gonna want to have them fire up their generative AI technology more than once to create additional avatars — at least when you have some down time. (See my last blog post, Generative AI Made Me Do It.)

After all, Lensa says every batch of headshots or selfies you send in will get you different results — even from the same set of photos! So, what the heck, when I had an idle moment during Christmas week, I dug out some additional shots of me, myself, and I, then hit the button to upload them. And, ta-dah, the results are below — which seemed to come through faster this time, by the way.

Note I did not submit any shots that had a Christmas or winter theme, but the smarty-pants AI at Lensa found a way to get into the spirit of the season. Pretty fun. (But then I’m easily amused.)

I hope this second post shows you more of what this amazing technology is capable of doing,  in this wild new realm of generative art. And just think, what will the next version of the technology behind this app (GPT-4) bring us when it comes out?

By the way, a belated Merry Christmas and a very Happy New Year to you!

Nine avatars of Graeme

Generative AI Made Me Do it

Having lived through countless cycles of artificial intelligence hype (I mean promise) over the last few decades, I naturally couldn’t avoid getting sucked into the latest hotness: generative AI. The attention this new flavor is getting lately is almost deafening. And that includes tons of startup funding and rich valuations in recent months, both of which have been declining in other sectors of the technology industry (to say the least).

With my insatiable curiosity for all things tech, I’ve been seeing and reading a lot about generative AI, both as it applies to written content and to art. (You can read some  articles I curate regularly on this topic here.) In particular, I was struck by what the latest version of the popular  Lensa app was doing — specifically, offering AI-generated avatars for anyone willing to sign up on its app and pay a nominal fee. You just submit 10-20 selfie or portrait shots of your mug, and — voila! — get back a whole bunch of AI-generated avatars. Most are what I’d call photo-like illustrations. Actually, it wasn’t all that “voila” — it takes a while for Lensa’s computers to generate these things. Hours for me. I suppose it greatly depends on how many people are signing up on a given day.

Anyway, most of the 200 that I got back were… well, the word crap comes to mind. But I pulled out six that were kinda nice. And, ta-dah, here they are — how generative AI sees Graeme. (Yes, this is an entirely self-absorbed post!) Continue reading

« Older posts Newer posts »