arrow

ACEC AI for AEC: Mastering Generative AI for Content Development

The following is a recap of my session for the American Council of Engineering Companies (ACEC) four-part series “AI for AEC.” This information is accurate as of September 2023.

The Rise of Generative AI

Artificial Intelligence (AI) has been part of our daily lives for a long time. Think Google Maps, predictive analytics, writing assistants like Grammarly, chatbots, Siri and Alexa, marketing automation and social platforms – just to name a few.

Yet, until very recently, we had to rely on software engineers and app developers to facilitate user access to AI platforms. In late 2022, ChatGPT went mainstream, creating universal access to the power of Generative AI through conversational chat interfaces that anyone can use.

This watershed moment in AI adoption caused an almost overnight explosion of hundreds of Generative AI tools. What took previous trailblazing technologies months and years to achieve, ChatGPT accomplished in days.

ChatGPT reaches 1 million users in 5 days

Top AI Uses for Content Development

With barriers to entry removed, everyone can use these tools! And everyone does or will be in a very short time, making these new tools “a must” to learn for your firm to stay competitive.

Don’t wait for the world to get smarter around you. (Paul Roetzer, Founder, Marketing AI Institute)

With more and more Gen AI tools entering the market almost daily, where do you begin with generative AI? How can you approach the rapidly evolving and seemingly limitless AI toolkit in a way that’s practical?

Don’t begin with tools and apps. Begin with what you’re trying to accomplish

Begin with what you’re trying to accomplish and what would help your firm bring greater efficiency and effectiveness into its daily content operations.

  1. Identify all content-related tasks. Create a list of all content development tasks and think about which of these tasks (especially repetitive, low-value tasks) could be easily automated, allowing your team to focus on more meaningful, creative work.
  2. Prioritize tools best suited to help with your specific needs. Many tools have overlapping features, and there are many one-point tools. Prioritize tools that are multifunctional and can help you with the most important or the most frequent activities.
  3. Review your existing marketing technology stack. Before you invest in something completely new, check what you already have. Salesforce, HubSpot, MailChimp, OpenAsset, Microsoft Office, Adobe Creative Suite, Canva, and most other technologies have already integrated AI features within their existing platforms.

With a clear picture of what you’re trying to accomplish, it will be easier to zoom in on the tools that could help you do that.

Finally, you have to allocate time – whether it’s an hour a day or an hour a week – to research, test and learn these new tools.

What Gen AI Can (and Can’t) Do

Generative AI is a powerful technology that can help your team accomplish more, faster. However, it’s important to understand what these tools are good for and where you need to rely on human experts.

What Gen AI Is Good For

AI is very good at research, brainstorming, production and productivity. Specific to content development and content marketing, AI tools can:

  • Help with audience or topic research and persona development
  • Quickly process and analyze your research data
  • Generate ideas and create outlines
  • Brainstorm headlines and subject lines
  • Suggest SEO optimization edits
  • Transcribe and translate
  • Create summaries, extract key points, rewrite and repurpose previously created content
  • Edit images, text, audio and video
  • Transform static content into more engaging and compelling pieces with the use of text-to-video or text-to-image tools
  • And so much more!

AI tools can also help write “functional content”– meeting summaries and takeaways, routine and repetitive email communications, organizing notes into outlines, creating descriptions and other similar types of content.

Creative or expertise-based writing is Gen AI’s weakest point. However, if you are not a natural writer or have a fear of a blank page, you could go as far as the first draft.

What Gen AI Is Terrible At

Don’t expect AI to offer original insights or a unique point of view – this is where critical thinking, expertise and empathy remain irreplaceable human skills.

Inherent Issues:

  • Data privacy and security
  • Misinformation (“hallucinations”)
  • Plagiarism
  • Copyright
  • Bias, ethics, and trust

Everything needs to be fact-checked, including any cited sources (ChatGPT has been known to make up sources that don’t exist).

If you work with sensitive or proprietary data, use open source models and run them on your own servers, ensuring that sensitive data isn’t leaking into the public domain. It will also give you more control for training and fine-tuning models on data and information specific to your firm, resulting in far superior output.

Prompt Engineering

With all Gen AI models, to complete any task, you have to master the art and science of prompting. And although you don’t need a degree in prompt engineering, you have to understand the basics and practice to improve.

At the core of prompting is the ability to provide good instructions and knowing the format in which to provide them. You have to use specific and relevant words to describe what you’re looking for – otherwise, the output will be very generic.

The better your ask is, the better quality information you’ll get in return.

There are many cheat sheets available, like this one from TrustInsights that’s free to download.

And, of course, there’s an AI app for that!

PromptPerfect

This could be a good playground to practice your prompting skills and get a bit of tutoring.

  • Optimizes prompts for different AI models
  • Provides “original” vs. “improved” comparisons

If you are working in teams, or multiple employees and multiple teams at your firm create prompts for similar tasks, build a prompt library that you can share across your firm. It will streamline content generation and create consistency in the output specific to your firm.

My Shortlist of AI Tools for AEC Firms to Consider

These are some of the AI content tools I’ve tested and liked. Some of them are now in my and my team’s standard workflows.

Note: Most Gen AI tools have user guides that can help you get started. There are also volumes of articles and tutorials. You don’t have to struggle all on your own!

Text

  • ChatGPT
  • Claude
  • Google Bard
  • Bing AI
  • Jasper
  • Writer
  • Copy.ai
  • Writesonic

Images

  • DALL·E
  • Midjourney
  • Stable Diffusion XL
  • Adobe Firefly (Beta)
  • Canva (AI features)
  • Adobe Photoshop (AI features)
  • Beautiful.ai (Presentations)

Video

  • Descript
  • Runway
  • Vidyo
  • VEED
  • HeyGen
  • Synthesia

Productivity

  • Grammarly
  • Otter.ai
  • Krisp
  • Motion

Select AI Writing and Editing Tools Comparison

ChatGPT

If you had to start with a single tool, this is the foundational tool you should have in your toolbox. There’s a free limited GPT 3.5 option. However, if you plan to use this tool in your regular workflows, I recommend upgrading.

Paid “Plus” subscription – $20/month:

  • Superior quality output
  • Internet access
  • Access to beta features
  • Extended capabilities through plugins and browser extensions

Claude 2

Claude from Anthropic can work with longer text than ChatGPT, so if you work with large volumes of text or multiple long-form content sources, you will need to use Claude.

There’s a free Claude 2 (beta) version, and you can upgrade to “Pro” for $20/month:

  • Can handle large volumes of text, faster
  • Great for summaries (long transcripts, multiple source documents)
  • Seems more intelligent and accurate than ChatGPT (has more updated data)

I also find its output often more sophisticated, so if I am not happy with what I am getting from ChatGPT, I turn to Claude for a “second opinion.”

Jasper

A more advanced content marketing tool and also more expensive – starting at $39/month per user (discounted is paid annually).

  • Trained not only on your brand voice/style but also key facts about your business, so the output is more tailored
  • Offers API and can be used everywhere with a browser extension
  • Has an AI image generator built in

Writer

Writer is built for enterprise-size teams. If you have a large marketing team or generate a lot of content and need to have tighter governance over the brand voice, Writer has these features built in.

It offers a free trial, after which it’s $18 per user/month (or $162 if paid annually):

  • Real-time brand style enforcement, including specific language choices important to your firm
  • Connect to your company data sources so you can use your databases, wiki and chats to create content
  • Content templates and reusable “snippets”
  • Enterprise API, self-hosted LLM

Copy.ai

Powered by OpenAI, it’s free for up to 2000 words/month or $49/month for unlimited words.

  • Upload and store your own data in “infobases” to reference it when creating new content
  • Brand style enforcement
  • Prompt improver
  • Built-in templates

Select AI Image Generation and Editing Tools

As a designer, I believe that image generation is still in its infancy, and at this moment in time, Gen AI tools perform better as image editors than image generators.

Overall, the results vary based on these three factors:

  1. The image purpose and type
  2.  The tool used
  3. The user’s ability to prompt

Adobe Firefly

Currently in beta, it has great promise but lags a bit behind other similar apps. Once it’s fully implemented, I believe it will become THE number one choice for millions of professionals who already have Adobe apps built into our daily workflows.

There’s also a growing list of AI tools from Firefly that have been integrated into Photoshop, as well as Illustrator and InDesign.

Firefly offers a free beta online, and some of the tools can be accessed via Creative Cloud beta apps.

  • Text to image generation
  • Generative fill (available in Photoshop beta)

DALL·E

DALLE from OpenAI (the same company that brought us ChatGPT) is the first AI-powered image generator.

All early adopters still get 40 free credits a month. Although DALL·E no longer offers free trials or free credits to new users, it’s only $15 for 115 credits.

  • It has both image generation and image editing capabilities
  • Its editing tools are in beta and much more limited than Stable Diffusion, for example
  • DALL·E is very intuitive to use, but it also produces pretty basic results

Midjourney

Midjourney is one of the more sophisticated image generators and produces probably the most realistic results.

It starts at $10/month for the Basic Plan, which gives you ~200 images and provides commercial usage rights.

  • You can only use it through Discord
  • There is some complexity to it as compared to DALL·E and some others
  • It has excellent tutorials and guides

Stable Diffusion XL

Stable Diffusion from Stability.ai can be accessed through two interfaces:

  • Dream Studio offers a lot of control over image creation, and results could be quite sophisticated. It starts at $10 for 1,000 credits.
  • Clipdrop offers both image creation and editing and has a free limited version and a Pro plan for $7/user per month with access to more advanced features and higher resolution images.

Clipdrop

The text-to-image prompt field has many different filtering options for image style and size and offers a negative prompt filter to specify what your image should not be.

Clipdrop Stable Diffusion AI image filters

There are currently nine editing tools, and these are my favorites:

AI for AEC Stable Diffusion Clipdrop tools

Uncrop: Expand your image horizontally or vertically (e.g., transform a horizontal project photo into a vertical cover image).

Remove and Replace Background: Remove background and/or create a different setting for your object.

Image Upscaler:

  • Scale up images up to 16x without losing quality (use an iPhone image as a full-bleed proposal cover)
  • Improve image quality by removing compression “grunge” and or sharpening the original (use scaled and compressed web-resolution JPEG for a full-bleed proposal cover)

AI image generation Stable Diffusion Clipdrop tools

Sample Workflow

Although I advocate for selecting tools that are multifunctional, as you integrate AI into your daily workflows, develop your own stacks for various marketing activities.

ACEC AI for AEC AI Tools for Content Development

For example:

  • I take a video of my presentation and use Descript to create a transcript of that video.
  • Depending on the length, I place that transcript into either Claude or ChatGPT with a prompt to create a summary with 3-5 takeaways.
  • I also ask Claude to brainstorm a dozen headlines. Although I would not use any of them word-for-word, I will likely be able to take inspiration from a few to come up with a good headline for my post.
  • I would edit the summary and use it as a description for the video and a LinkedIn post. I will also be able to use the takeaway bullets to create short social media captions.

Resources

I hope you are excited to test the new tools, sharpen your prompting skills and make generative AI work for you.

To help you keep tabs on new AI tools, trends and broader implications of AI on all areas of business, below are a few of my favorite resources.

Ben’s Bites is a daily digest of what’s new in the world of AI.

Marketing AI Institute offers free resources/webinars and paid courses specifically tailored to AI for marketing.

Pay attention to research and content published by large global management consulting firms:

  • Deloitte AI Institute
  • West Monroe
  • McKinsey & Company

Please don’t hesitate to reach out to me via email or LinkedIn if you have any questions or would like additional information.


Actionable Ideas Straight to Your Inbox

Don’t miss another article! Sign up to receive expert advice to help you powerhouse your branding and marketing and stay ahead of your competition – delivered monthly to your inbox.

Subscribe

4463