ech became a chatbot. Here’s what we have learned so far. //

OpenAI’s ChatGPT release got everyone, including ourselves, thinking about its impact. We began to ask big questions and wonder about the impact of ChatGPT, as well as other technologies, on marketing, websites, and customer experience. We asked:

• Will this type generative AI replace websites one day?
• Can this technology be helpful to marketers?
• What are some emerging use cases that we should investigate?

The MarTech community was able to embrace the changes and opportunities that generative AI offers.

We decided to build our own bot. John Blanchard, senior director of our marketing technology department, and I built a bot. We created an index of the MarTech.org posts and pages, and then embedded them in MarTechBot. This sits on ChatGPT.

Here are some initial lessons we’ve learned:

Humans are known to be destructive.

Humans are prone to breaking things (perhaps to confirm their biases?) MarTechBot is not perfect or always right, so it’s not very good. Many users have told us this. The results are a bit erratic depending on the input you enter. MarTechBot has had hundreds of conversations, many of which were multi-question threads. In one chat, for example, the user asked two questions in succession and received answers to both.

• What are the main considerations we should make before conducting a marketing technology audit in its entirety?
• What are the most important questions to ask internal stakeholders?

In addition, we have discovered a bug which causes a “token” error. We know the reason for it but cannot fix it yet. Stay tuned!

Nobody seems to understand how this works

This is brand new technology. It’s new tech.

Although it feels like one, this is not a search tool.

Baratunde Thuston said on the Pivot Podcast that using these tools can make you feel as if you’re casting spells. He’s right. The tools of generative AI are becoming more and more like “mechs”, giving me capabilities and powers I’m still discovering and learning how to use.

These powers are not without their limitations. We don’t really understand the reasoning behind the bots and language models. Great power brings great responsibility. (And the requirement to carefully read and double-check your work! ).

MMM, what?

Here’s an example. MarTechBot replied to a user who asked “Tell Me What I Need to Know About MMM?” with the following information:

AI: Marketing Mix Modeling is a technique that marketers use to measure the impact advertising and promotion across channels, while controlling external factors such as inflation and consumer sentiment. MMM’s goal is to assist marketing leaders in planning future marketing expenditure and measuring the performance of previous investments. MMM can serve three purposes: it can be used as a scorekeeper to track the incremental impact of marketing on the business, as a forecaster to determine the effect that increasing or decreasing marketing budgets will make on marketing’s overall contribution, and as a coaching tool to suggest improvements to existing marketing investments. MMM offers a significant opportunity to improve returns on media investment, but models must be reevaluated regularly to deliver expected results.

The same query as in ChatGPT was

MMM can mean different things depending on context. Here are three possible meanings:

MMM is a financial acronym:

MMM is the ticker symbol of 3M Company. This multinational conglomerate produces a variety of products including medical products and office supplies. The company is listed at the New York Stock Exchange.

MMM is a cryptocurrency.

MMM Global, as it describes itself, is a “community” of people who provide each other with financial assistance on the basis of reciprocity and benevolence. It was founded by Sergei Mavrodi in 2011, a Russian entrepreneur, and operates allegedly as a peer-to-peer decentralized network. MMM Global has its own cryptocurrency called MMM Coin that it uses to exchange funds within the community.

MMM Global is a company that has been involved in multiple Ponzi and pyramid schemes over the past few years. They have also faced legal actions in many countries.

MMM is a common texting abbreviation.

MMM is an abbreviation of “mmm-hmm,” a way to indicate agreement or acknowledgement. MMM can be used to express appreciation or pleasure, such as “MMM, that pizza is delicious!”

Bard has given me this result:

MMM is the 3M Company. The American multinational conglomerate is headquartered in Maplewood Minnesota. The company has a diverse technology portfolio that includes consumer, industrial and safety products. 3M employs over 90,000 people and is active in more than 70 different countries.

Here are some other interesting facts about 3M. You get the idea.

Is it true that MarTechBot has been “pre-prompted”, since we have trained it with MarTech.org’s content? I believe so. It’s amazing to think about the implications of context and prompting. MarTechBot’s response makes sense, but there was no prompt. Just the letters MMM. Wild.

Many people are still unsure of how to use chatbots that generate content.

We don’t know how to use them, and we aren’t even sure of their purpose.

Multi-layered learning seems to be the best way to interact. It is easy to type “SEO strategy”, but it’s not nuanced. Specificity is more valuable. Give me a SEO strategy for a flower shop in Portland. We run a small influencer programme and blog regularly”, which is likely to yield better results. By asking follow-ups questions, you can refine and fine tune results. It’s not just a search engine. It’s different. Alien almost.

The more I try, the further down the rabbit hole I go. Yesterday, I created a quiz to test my SEO knowledge. I used today’s prompt to improve my sales skills by simulating an interaction with the CMO of a national company who I was trying convince to beta-test my fictional analytics software.

This has incredible implications. Bill Gates may be on to something when he says AI will replace teachers.

As humans become more comfortable with these tools (and learn how to live with them), new applications will appear. Earlier, I described an idea I had for creating a quiz that would simulate a sales pitch and self-test. I created a simple Javascript Calculator using Google Bard today to help me consider different email scenarios. Then, I created an interactive checklist for podcasting.

There are many possibilities.

What next?

MarTech.org is a great resource, and we wanted to make sure that people knew about it. It’s now a collaborative effort with the MarTech community. The feedback has been mixed. Some people are happy that we launched MarTechBot and others are disappointed by the token bug, or the bot’s slowness compared to ChatGPT. We learn from every comment. What’s the joy of beta-launching something?

We are looking for ways to improve this bot, and we want your feedback. Try MarTechBot and you’ll find the feedback form at the bottom of that page. I am looking forward to hearing from your.

You can join me on LinkedIn for real-time chat and information.

This is what we’ve learned (so far)./a> appeared first on a href=”https://martech.org” rel=”nofollow”>MarTech/a>. The post This is what ). has taught us so far appeared first on MarTech.