LLM Brain Rot and You

It's been a while since I last posted on AI, LLMs, and chat bots. Here's what I've been up to. I use Gemini to help me write business letters, and at work I use ChatGPT as a SQL and AMPScript code assist tool. ChatGPT is decent for helping a non software developer learn the quirks, in's, and out's of programming languages and scripts. It has not been smooth sailing. Each day I use ChatGPT with version history enabled, it seems to forget what it just said in its last response to a question or prompt and automatically starts correcting itself if you provide it with its own SQL query to make updates to.

The hallucinations are by far the worst bug in the LLM. ChatGPT proactively adds field names and table aliases that don't exist in the original query or prompt. It also uses SQL commands that aren't available in SFMC's version of SQL. And, after I have already specified that the query needs to work with SFMC's SQL (based on T-SQL), it gives me SQL that I already know can't run in SFMC. I have to remind it each time it does this with an additional prompt that the SQL can't start with "with" before "select". 

Anyhow, even with its flaws, dev time has been greatly reduced from a few days to a few hours in most cases.

The prompt you give it is only as good as your understanding of how a tool or process works. Using an LLM is a lot like using a mobile device app's GPS. Sure, the GPS gets you to your location with turn by turn directions, but it doesn't teach you how to navigate a map by yourself. How many times have you heard people say, oh I don't know where [location] is, I just ask the maps app to get me there?

It won't help you think critically or write professionally unless you ask it how it analyses text or data before giving you a response. Even then, it seems bound by programmed how to.

Supreme Court Justices, reimagined by Bing AI as D&D characters - part 1

(obligatory attribution text) This content was generated by Microsoft Bing search chat mode, a conversational interface that can answer questions, generate creative content, and provide suggestions for the next user turn. To learn more about Bing chat mode, visit https://www.bing.com/chat.

Here is a possible table of D&D stats, alignment, class, and guild for the current 9 justices:

Name	Stats	Alignment	Class	Guild
John G. Roberts	STR 12 (+1), DEX 14 (+2), CON 14 (+2), INT 18 (+4), WIS 16 (+3), CHA 16 (+3)	Lawful Neutral	Bard	Order of the Harpers
Clarence Thomas	STR 12 (+1), DEX 18 (+4), CON 16 (+3), INT 16 (+3), WIS 20 (+5), CHA 12 (+1)	Lawful Neutral	Monk	Way of Tranquility
Samuel A. Alito	STR 14 (+2), DEX 10 (0), CON 16 (+3), INT 16 (+3), WIS 14 (+2), CHA 14 (+2)	Lawful Evil	Fighter	Battlemaster
Sonia Sotomayor	STR 10 (0), DEX 12 (+1), CON 14 (+2), INT 18 (+4), WIS 18 (+4), CHA 16 (+3)	Neutral Good	Wizard	School of Divination
Elena Kagan	STR 10 (0), DEX 14 (+2), CON 14 (+2), INT 20 (+5), WIS 16 (+3), CHA 18 (+4)	Neutral Good	Bard	College of Lore

Brett M. Kavanaugh	STR 12 (+1), DEX 10 (0), CON 14 (+2), INT 16 (+3), WIS 12 (+1), CHA 14 (+2)	Lawful Evil	Rogue	Assassin
Amy Coney Barrett	STR 10 (0), DEX 14 (+2), CON 12 (+1), INT 18 (+4), WIS 16 (+3), CHA 16 (+3)	Lawful Good	Cleric	Order Domain
Ketanji Brown Jackson	STR 10 (0), DEX 16 (+3), CON 14 (+2), INT 18 (+4), WIS 18 (+4), CHA 16 (+3)	Neutral Good	Druid	Circle of the Land

Prompt: short skit explaining iterative analytics

Sure, here is a short acting skit between two marketing professionals explaining iterative analytics:

---

Scene: A conference room at a marketing agency.

Characters:

• Ashley: A marketing manager
• Ben: A data analyst

---

Ashley: So, Ben, I'm really excited about this new iterative analytics approach we're trying.

Ben: Me too, Ashley. I think it has the potential to really revolutionize the way we do marketing.

Ashley: So, can you explain to me how it works?

Ben: Sure. Basically, it's a process of continuously testing and iterating on our marketing campaigns based on data insights.

Ashley: So, we start with a hypothesis, and then we test it with data. If the data doesn't support our hypothesis, we iterate on the campaign and try again.

Ben: Exactly. And we keep doing this until we find a campaign that's really successful.

Ashley: That sounds great. So, what are some of the benefits of iterative analytics?

Today's parlay with Bard

Q: What can you be proactive about if a human doesn't give you commands

Bard AI:

There are a few things that I can be proactive about if a human doesn't give me commands.

I can try to anticipate their needs. For example, if they are working on a document, I can suggest relevant sources or tools.

I can also try to learn their preferences. For example, if they prefer to work in a certain way, I can try to adapt to their workflow.

Finally, I can always ask for clarification if I am not sure what they want me to do.

I am still under development, but I am learning new things every day. I am always looking for ways to be more helpful and proactive.

Generative AI for Marketing

Bard AI says...

Generative AI is a powerful tool that can be used to create a variety of marketing materials, including blog posts, social media posts, and even ad copy. However, there are some pros and cons to using generative AI for marketing that you should be aware of.

Pros of using generative AI for marketing:

• Increased efficiency: Generative AI can help you to create more marketing materials in less time. This can free up your time to focus on other aspects of your marketing campaign.
• Improved personalization: Generative AI can be used to personalize your marketing materials for individual users. This can help you to connect with your audience on a more personal level and increase engagement.
• New and creative ideas: Generative AI can help you to come up with new and creative marketing ideas. This can help you to stand out from the competition and generate more leads and sales.

Cons of using generative AI for marketing:

• Potential for bias: Generative AI is trained on a massive amount of data, which can introduce bias into the generated content. This is something that you need to be aware of and take steps to mitigate.
• Quality of content: The quality of the content generated by generative AI can vary. It is important to proofread and edit the content before you publish it.
• Cost: Generative AI can be expensive to use. You need to factor in the cost of the software, as well as the cost of training and maintaining the AI model.

(human here)

When I asked Bard how generative AI was unique, it said that generative AI was more efficient than humans and could produce content at scale. 

Artificial Intelligence in Human Resource Management

That's the theme of this month's online conference from HR.com. If you thought the human part of human resources was broken for many companies, could AI be making it a better or worse experience for candidates, employers, and potential partners? 

Hello Alexa, part 2

It's been a while since I last posted about Amazon's Alexa being accessible through the Amazon retail shopping app. And while Amazon's generic web search is fine and comes nowhere near the voice-to-search recognition that Google search offers, Amazon is missing the point about monetizing the index system that they have for the millions of products listed on their shopping exchange.

Wouldn't it be better if instead of matching to keywords (mostly nouns) in a user's speech search, that Amazon served up relevant recommendations instead.

Say for example, you ask Alexa (in the Amazon app):

"recommended wines for dinner" or 
"recommended red wines" or 
"recommended fruity wines"

Alexa currently offers no recommended product searches for any of the wines or wineries that sell on Amazon. Well, it certainly can't recommend wines that's for sure. But it could if Amazon incorporated product label text, certified wine reviews, or wine manufacturer descriptions in what can be searched. The words "recommend" and "recommended" are not in Alexa's lexicon of search knowledge. Perhaps this is too advanced a concept for Amazon's AI.

You can still just say "red wine" or "white wine" and those options will show up with valid results in the Amazon app.

Voice searching the Amazon product engine should be no different than typing in the search query. 

The results are mixed, however.

You can say "services for window washing near me" and Amazon's app will show for "Hire a Window Cleaner" (Amazon Home Services) as the top result. That's spot on. The third result (same screen on a smartphone) shows "Window Cleaning" (Amazon Home Services), also a valid result to what I was voice searching for.

Maybe this is a phased rollout for voice search queries.

Deep Learning

Deep learning has its origins in the early days of artificial intelligence, when researchers began to explore the use of artificial neural networks to learn from data. However, it wasn't until the early 2000s that deep learning began to gain popularity as a field of study. This was due in part to the development of new algorithms that made it possible to train deep neural networks on large datasets. Additionally, the availability of high-performance computing resources made it possible to train deep neural networks in a reasonable amount of time.

In 2012, Geoffrey Hinton and his team at the University of Toronto used deep learning to achieve a breakthrough in image recognition. Their algorithm, called AlexNet, won the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) by a significant margin. This victory helped to spark a renewed interest in deep learning, and the field has since exploded in popularity.

Today, deep learning is used in a wide variety of applications, including image recognition, natural language processing, speech recognition, and machine translation. Deep learning is also being used to develop new drugs, create self-driving cars, and improve the accuracy of weather forecasts.

As deep learning continues to develop, it is likely to have a major impact on a wide range of industries. It is already being used to solve some of the world's most challenging problems, and it is only going to become more powerful in the years to come.

Hello Alexa

Not sure how long ago this feature was added, but it looks like someone just replaced the default microphone app with Alexa's voice and mannerisms on the Amazon app. You'd think that if you were accessing Alexa from within Amazon's shopping app, that the default search would be for items listed in Amazon's eCommerce ecosystem. Sadly, this is not the case.

Screenshot of Alexa's Intro Screen on Amazon App

My first query: "weather tracking for the home", followed by "weather tracking apps"

I don't like Alexa's color bar acknowledgement followed by its electronic beep. For the few seconds it takes to execute these robotic response commands, it is an unnecessary feature. Alexa responds by verbally giving me the weather forecast for Salem Oregon.

The response is puzzling because I was just adding/removing items from my wish lists in the app which one could assume that I am already logged into my account which has my mailing address in it (and I don't live in Oregon). Even if location services were turned on for this app, surely the developers would have programmed that into Alexa -- to be able to give regional information based on already known criteria.

My next query: "search Amazon for home weather tracking"

That brought up a relevant search list on Amazon's store.

Artificial Intelligence is only as good as the team that builds it.

I can just visualize the disconnect between the business user story and what got implemented by the development team. Maybe I'm just disappointed because I'm so used to Google search providing accurate, relevant results from text or voice queries.

At least Alexa can tell jokes (Siri cannot):

"Tell me a funny cat joke"

Alexa: What does a cat say when it gets hurt? Me-ow.