GrowthPath Partners LLC

Empowering Purpose-Driven Growth

Liza Adams

GPT-4o Images: Impressive, But Spelling Still Stumbles

· ·

I’ve never heard of the “Assohly Line” of 1913 and “Orfice Work” in the 40s-60s. 🤪

But that’s what showed up in the timeline image created by the new “Images in ChatGPT,” announced this week as part of GPT-4o.

I fed it its own deep research on 100 years of job evolution in the U.S. and asked it to turn the insights into a timeline infographic.

The layout was clean. The flow made sense. The icons and structure were well done.

It’s definitely better and genuinely impressive, but still not quite there with text placement and spelling. It struggles with small labels, dense copy, long words, and text-heavy infographics. And OpenAI has acknowledged this limitation.

That said, I’ve seen others share infographics that turned out great using this new feature. So results may vary, but there’s no harm in trying it for yourself.

Where GPT-4o image generation works well:

  • ► Posters and diagrams

  • ► Concepts with lighter or no text

  • ► Visuals for storytelling

This isn’t a complaint, just a reminder of how far things have come in such a short time. We had nothing close to this two and a half years ago.

And remember: This is the least capable AI image generation we’ll see from here on out.

Try it yourself and let me know what you think.

BTW, “Assohly Line” and “Orfice Work” are supposed to be Assembly Line and Office Work. 😉

(Here’s the same concept generated by Claude, another AI assistant, for comparison: https://lnkd.in/gSNPwyRR)

#AIVisualization #AIImages #ChatGPT #Infographics #AIDesign

See original post here

Don’t Fact-Check ChatGPT: Use It as a Thought Partner

· ·

Did you know that depending on the AI model, ChatGPT can confidently give WRONG answers (hallucinate) 37% to 80% of the time?

That’s not a flaw. (Heck, I think I gave 37% to 80% wrong answers in my electromagnetics class in engineering school. 🤪) It’s a sign we might be misunderstanding what ChatGPT is actually best at.

AI tools like ChatGPT aren’t designed to be fact-checkers or Q&A machines. They’re built to be your thought partner, idea generator, and brainstorming buddy.

(Here are ways and example of using AI as a thought partner: https://lnkd.in/gSk7wj-Y)

In fact, sometimes the best thing ChatGPT does isn’t give you the answer. It’s asking the right question back.

For example:

  • If you give it a vague goal, it might ask:
    “What outcome are you hoping for?”

  • If you’re drafting a message or a piece of content, it might ask:
    “Who’s the audience for this?”

  • If you’re stuck on a decision, it might ask:
    “What would happen if you did nothing?”

These kinds of questions help you slow down, get clarity, and think in new ways. The simple chat box makes it easy to assume ChatGPT is just for questions and answers like a smarter search engine. But it’s actually much better at helping you think than helping you win Jeopardy.

If you’ve never (or rarely) noticed ChatGPT giving wrong answers, you’re probably using it exactly right.

Wrong answers aren’t always a bad thing. They can help with:

  • What-if thinking

  • Fresh ideas

  • Strategy brainstorming

  • New perspectives

The key is giving it good context, clear prompts, and knowing what it’s for. Connecting it to outside sources (like web search) can help when facts are important.

Here’s a simple test: If wrong answers bother you, you might be using ChatGPT like a fact-checker instead of as a thought partner.

Change how you use it. Try it this way, and you’ll get better results.

#AIHallucinations #AICollaboration #ThoughtPartner #ChatGPT #AIModels

See original post here

Don’t Ask ChatGPT for Facts, Ask It to Think

· ·

Published on 2025-03-26 13:31

Did you know that depending on the AI model, ChatGPT can confidently give WRONG answers (hallucinate) 37% to 80% of the time?

That’s not a flaw. (Heck, I think I gave 37% to 80% wrong answers in my electromagnetics class in engineering school. 🤪) It’s a sign we might be misunderstanding what ChatGPT is actually best at.

AI tools like ChatGPT aren’t designed to be fact-checkers or Q&A machines. They’re built to be your thought partner, idea generator, and brainstorming buddy.

(Here are ways and example of using AI as a thought partner: https://lnkd.in/gSk7wj-Y)

In fact, sometimes the best thing ChatGPT does isn’t give you the answer. It’s asking the right question back.

For example:

  • If you give it a vague goal, it might ask: “What outcome are you hoping for?”

  • If you’re drafting a message or a piece of content, it might ask: “Who’s the audience for this?”

  • If you’re stuck on a decision, it might ask: “What would happen if you did nothing?”

These kinds of questions help you slow down, get clarity, and think in new ways.

The simple chat box makes it easy to assume ChatGPT is just for questions and answers like a smarter search engine. But it’s actually much better at helping you think than helping you win Jeopardy.

If you’ve never (or rarely) noticed ChatGPT giving wrong answers, you’re probably using it exactly right.

Wrong answers aren’t always a bad thing. They can help with:

  • What-if thinking

  • Fresh ideas

  • Strategy brainstorming

  • New perspectives

The key is giving it good context, clear prompts, and knowing what it’s for. Connecting it to outside sources (like web search) can help when facts are important.

Here’s a simple test: If wrong answers bother you, you might be using ChatGPT like a fact-checker instead of as a thought partner.

Change how you use it. Try it this way, and you’ll get better results.

#AIHallucinations #AICollaboration #ThoughtPartner #ChatGPT #AIModels

The Future of Marketing: 9 Roles Reshaped by AI

· ·

Practical AI in Go-to-Market
Get practical insights in using AI for go-to-market strategy, initiatives, workflows, and roles.

Most marketers are asking which AI tools to use. But the more useful question is what your role will look like once you’re using them.

Previously, I shared a visual showing how the Product Marketing role might evolve with AI.

Based on the response, it’s clear that many of us are ready to talk about what’s next for our jobs and careers.

So with the help of AI, I mapped out how nine marketing roles might change. This includes Content, Brand, Growth, Digital, Field, Demand Gen, Partner, Operations, and Product Marketing.

Each slide in the carousel below shows how one role might move through three phases:

  • Using AI tools

  • Guiding AI teammates

  • Orchestrating AI systems

Not every team or function will move through all three. And you don’t have to pick just one phase. Most companies will likely have a mix and the mix will change over time.

I believe the projections are reasonable. In fact, I’m already seeing some strong signals based on how various marketing teams I work with use AI.

Some of these roles may merge together as well, not just across workflows but also across teams and orgs.

This builds on the framework I shared in last week’s newsletter:

AI is Redefining GTM Jobs: From Tool Users to Teammates to Orchestrators

And continues the conversation started here: https://lnkd.in/gbCxHws3

Swipe through. See where your role might be headed. Would love to hear your take. Are you and your teams upskilling to step into these emerging and future roles?

#FutureOfWork #AIJobs #AIAdoption #AILiteracy #AIinMarketing

Check out this evolution yourself using the interactive models below for various roles in marketing, sales, and customer success:

1) Interactive Marketing Role Evolution Model – https://claude.site/artifacts/4ef7e0ed-bc24-4664-9c92-ba4330981c43

2) Interactive Sales Role Evolution Model – https://claude.site/artifacts/1da805a7-f39a-42ba-9fc0-3a3c60c1344d

3) Interactive Customer Success Role Evolution Model – https://claude.site/artifacts/949f0b33-2f2f-4be1-8b63-f9b4807efa55

See original post here

How AI Will Transform Your Marketing Job

· ·

Most marketers are asking which AI tools to use. But the more useful question is what your role will look like once you’re using them.

Previously, I shared a visual showing how the Product Marketing role might evolve with AI.

Based on the response, it’s clear that many of us are ready to talk about what’s next for our jobs and careers.

So with the help of AI, I mapped out how nine marketing roles might change. This includes Content, Brand, Growth, Digital, Field, Demand Gen, Partner, Operations, and Product Marketing.

Each slide in the carousel below shows how one role might move through three phases:

  • Using AI tools

  • Guiding AI teammates

  • Orchestrating AI systems

Not every team or function will move through all three. And you don’t have to pick just one phase. Most companies will likely have a mix and the mix will change over time.

I believe the projections are reasonable. In fact, I’m already seeing some strong signals based on how various marketing teams I work with use AI.

Some of these roles may merge together as well, not just across workflows but also across teams and orgs.

This builds on the framework I shared in last week’s newsletter:
AI is Redefining GTM Jobs: From Tool Users to Teammates to Orchestrators, https://lnkd.in/gYz3DKhA

And continues the conversation started here: https://lnkd.in/gbCxHws3

Swipe through. See where your role might be headed. Would love to hear your take. Are you and your teams upskilling to step into these emerging and future roles?

#FutureOfWork #AIJobs #AIAdoption #AILiteracy #AIinMarketing

Check out this evolution yourself using the interactive models below for various roles in marketing, sales, and customer success:

1) Interactive Marketing Role Evolution Model – https://claude.site/artifacts/4ef7e0ed-bc24-4664-9c92-ba4330981c43

2) Interactive Sales Role Evolution Model – https://claude.site/artifacts/1da805a7-f39a-42ba-9fc0-3a3c60c1344d

3) Interactive Customer Success Role Evolution Model – https://claude.site/artifacts/949f0b33-2f2f-4be1-8b63-f9b4807efa55