Navigating the New Frontiers: Google's Search Generative Experience, MusicLM, and Microsoft's AI Skills Initiative

Welcome to Aideations. The most comprehensive daily AI newsletter on the planet! How do I know? Because I read over 50 of them, plus the news, so you don’t have to. It’s my goal to make this the most-read newsletter on AI. I can’t do that without your support and feedback.

Here's what we've got in store for you today:

🤿 Diving In To Google SGE

🎸 Testing Google MusicLM

👾 Become An AI Pro With Microsoft

🧠 Unraveling “Dark Patterns”

📚 Research Of The Day

🎥 Video Of The Day

🛠 Tools Of The Day

🤌 Prompt Of The Day

🐥 Tweet Of The Day

A Deep Dive Into Googles AI Powered SGE

Hold onto your marketing hats, folks. Google is at it again, tinkering with its search engine like a kid with a new Lego set. And if you're anything like me, you're probably downing your fifth cup of coffee, anxiously pondering over their latest creation: the AI-powered Search Generative Experience (SGE). I just gained access to test these news features and of course that put me into a rabbit hole of trying to figure out what all this means for us marketers.

I bet you're rattling off questions faster than an auctioneer on a sugar high: Will these new-fangled AI-generated responses kill off site clicks? How will my carefully crafted search ads fare in this brave new world?

I won't string you along; we're in somewhat uncharted waters here. The SGE experiment is an opt-in affair, so we'll be watching it evolve in real time. But there are a few tidbits we can squeeze out of Google's public announcements that might help quell your caffeine-fueled anxiety.

First off, Google isn't planning to hang you out to dry. They've reassured us that their mammoth search engine will remain focused on driving traffic to websites. They're also promising to keep search ads as a "native part of SGE."

So, let's roll up our sleeves and delve into this mysterious SGE beast and what it could mean for marketers and advertisers across the business spectrum.

SGE, in a nutshell, is Google's experimental playground where artificial intelligence gets to flex its muscles, providing contextual answers to your complex questions. Here's how it's supposed to enhance your search journey:

1. AI snapshot: This nifty feature crafts a unique summary for those long-winded queries you toss at it.

2. Conversational mode: SGE gets chatty, answering your follow-up questions while holding onto the context of the original search.

3. Vertical experiences: For commercial searches, it provides a tailored list of features and more product details.

Let's break these down, shall we?

AI Snapshot

Imagine this scenario: you're diving headfirst into a Google abyss, trying to figure out whether a beach vacation in Hawaii or a camping trip in Yellowstone is the best family-friendly vacation where Fido is also welcome. Usually, you'd be swamped with a flurry of links, each just touching on fragments of your complex question. But that's where SGE (Search Generative Experience), Google's latest brainchild, steps in. SGE stitches together info-nuggets from an array of sources to hand you a neat, concise answer, lovingly referred to as a "snapshot."

Consider it like Google's trusty response to even the quirkiest queries, such as "which is the best dog-friendly resort near a ski area that also has indoor waterpark facilities for kids?" Just like a reliable old friend, Google's got your back, even when your questions have more twists and turns than a roller coaster ride.

But hold up, SGE isn't planning to steal the show and leave web content out in the cold. Google's made it clear that its loyalties lie in directing users towards relevant online content. So, if you're a business betting on organic or paid search traffic, it's far from the endgame.

SGE's masterstroke? A trio of links, neatly positioned next to each snapshot, complete with eye-grabbing thumbnails. Plus, there's a "bear claw" (Google's cute, homegrown term) icon that opens up the snapshot, showing you the source of each sentence. It's like building a bridge of transparency and trust, a move that's pretty refreshing in the AI world, especially considering the dubious rep of tools like Bard.

Conversational Mode

Google knows your search journey isn't a one-question deal. That's where the conversational mode comes in. It's designed to answer follow-up questions, keeping the context of the original search intact.

During a live demo at Google I/O, Kathy Edwards showcased this feature by asking "why do whales like to sing," followed by "plush ones for kids under $40". A regular search engine would be clueless about her interest in plush whale toys, but not our SGE. It's a whiz at connecting the dots from previous searches.

The takeaway here is the opportunity for businesses. By optimizing content to align with this sort of conversational searching, there's a golden chance to draw more relevant traffic. So, it's time to double down on your high-quality online sources and enhance your E-EAT cred.

As for the paid side, it might be wise to flex those phrase and broad match keywords in your account to catch these unique long-tailed searches. And with these longer queries on the rise, regularly checking your search terms and adding new negative keywords is going to be as important as remembering to water your office plant.

Vertical Experiences

SGE also jazzes up vertical searches—think shopping and local business searches—especially when there are multiple aspects to your query. For businesses advertising and selling online, this is where it gets exciting.

Let's say you're bidding adieu to your car and want to cycle to work instead. But not any bike will do. You're looking for a trusty steed that can handle a five-mile, hilly commute. SGE gets this. It rummages through the 38 million products in Google's Shopping Graph and serves up the perfect result.

If you further refine your search, for instance, asking for red bikes that fit the bill, SGE steps it up. It's like having a personal shopper on standby. And for advertisers, SGE's increased use of visuals might signal the need to incorporate image extensions with your ads sooner rather than later.

SGE Implications for Businesses

For all its bells and whistles, SGE also raises some concerns. Google's statement on the ads experience in SGE was decidedly vague, leaving us hanging like a cliffhanger season finale. But, here's what we've gathered so far:

Ad Placement: SGE results pages will have dedicated ad slots, and while they're labeled as "Sponsored", they're from existing campaigns, with no option for advertisers to opt-out at present.

Measurement Black Box: As it stands, there's no way to see how ads on SGE SERPs perform separate from regular searches. You can't opt out of having your ads show up on SGE results, and all the data is bundled together like a mystery meat sandwich.

Potential Organic Traffic Impact: The AI snapshot could swallow up most of a mobile screen, pushing the familiar 10 blue links down the page and possibly throttling organic traffic. Plus, the uncertainty on how to rank for these new features adds to the woes.

Yet, it's not all doom and gloom. If SGE can do the "heavy lifting" and qualify potential shoppers before they reach your site, you could be looking at higher conversion rates.

In terms of preparing for this brave new world, my advice is to sign up for Search Labs and opt in to the SGE waitlist. If you notice a dramatic shift in your campaign results, you'll at least be able to see if it's a keyword getting the SGE treatment.

In short, SGE is the wild west of search experiences, and we're all in for a riveting ride. So, buckle up and keep your marketing wits about you. After all, who doesn't love a good adventure, right?

Google MusicLM Is Cool & Promising But Not Extremely Impressive, Yet.

Your boy got access to the cool newish tool on the block for producing music using nothing but text. Let’s just say I was extremely excited to try this out.

Since it only generates 19 seconds of audio and two tracks per prompt, its application is very limited. Now could I see an artist using this to brainstorm or sample a potential new melody for a track? Absolutely!

However, the quality just isn’t there yet. But just like every other research project, things tend to move pretty quickly. We went from Midjourney just being kinda gimmicky to producing images even the most well-trained eyes and photographers can’t tell whether it’s real or AI-generated. Not to mention AI artists have won several awards when compared to human artists and fooled judges.

The music that MusicLM produces is more of an idea than something you could actually use on a track. I did find it easier to produce electronic music, coincidence? I think not.

Either way, I think the future of music is bright, and more and more creators will have access to these types of tools will create more music than we ever thought possible. The future may just be producing the next hit song with a few well-crafted prompts.

Becoming an AI Pro: Microsoft's New Initiative Turns the Tables

The AI genie isn't just out of the bottle—it's on a rocket headed for the moon. Microsoft, in its infinite tech wisdom, is now offering a front-row seat to this high-speed ride. They're launching the AI Skills Initiative, designed to turn everyday Joes and Janes into AI pros. Just think of it as your private, crash-course tutorial for generative AI. You can jump in HERE, I’ll race you. I’m kidding, I spent last night and this morning earning my certifications.

But what's the course about, you ask? Well, initially served in plain old English, the program dishes out everything from the ABCs of AI to the ethical frameworks for its use. But the cherry on top? A shiny certificate, the perfect new addition to your LinkedIn (yes, Microsoft owns it too) profile. And if English isn’t your cup of tea, they’re promising more linguistic flavors—Spanish, Portuguese, French, German, Simplified Chinese, and Japanese—all "over the coming months". Nothing like a good, global language buffet, eh?

Wait, there’s more! Microsoft's not leaving educators behind. A toolkit specifically designed for educators is also in the pipeline, plus a separate AI course for teachers and trainers. Microsoft’s ensuring nobody gets left behind in this AI-powered sprint.

Now, for the grand finale, Microsoft's upping their AI game with the Generative AI Skills Grant Challenge. In collaboration with data.org, Microsoft’s AI for Good Lab, and GitHub, they’re championing a cause for historically marginalized populations. The lucky grant recipients will get a swag bag of goodies, from financial support to group learning opportunities and access to Microsoft events and cloud resources. You've got until August 15th to toss your hat in the ring. This initiative from Microsoft is a glaring neon sign that AI isn’t just a sci-fi dream—it’s here, it’s now, and it’s got a spot reserved for you.

Caught in the Subscription Web: Unraveling 'Dark Patterns' and AI Tricks

Last time I checked my credit card statement, I felt like I was decoding a hieroglyphic with all the forgotten subscriptions popping up. Don't feel singled out if you're in the same boat. The USA was on a subscription sign-up bender last year, but now, cancellations are taking the lead, making leaving feel like an escape room game. When I tried to bid adieu to Amazon’s Audible, they tempted me with a “pause” button instead of losing unused book credits. I bit, and now have a friendly reminder set for October.

The sneaky online sales techniques—fondly called “dark patterns”—have been in the regulatory spotlight for a while now. It's like companies have turned into digital Jedis, using AI, data mining, and algorithms to keep customers hooked. The US Federal Trade Commission (FTC) isn't playing along though, recently suing Amazon for allegedly trapping people into recurring Prime subscriptions. They even made Publishers Clearing House cough up $18.5mn for enticing customers into purchases while entering sweepstakes.

As 'dark pattern' crackdowns escalate, companies are betting on AI sophistication. Imagine a sales AI that knows not just what you're likely to buy, but when you're most likely to buy it. That could even anticipate your emotional state and strike when you're most vulnerable. It's like a creepy salesperson who knows just a bit too much about your life.

Picture asking your AI for robot mop info, only to be led to a pre-filled sign-up page for a one-year subscription. You shrug and click 'yes', missing out on a potential free trial. According to Matthias Holweg of Oxford’s Saïd Business School, as AI gets craftier, we need regulations to prevent us from being puppeteered.

The fight against potentially manipulative sales practices is ongoing. The FCA’s new consumer duty will specifically caution companies against exploiting customer biases. The European Parliament is crafting legislation to set boundaries for AI technologies. Meanwhile, the FTC wants companies to maintain records of their psychological and behavioural research. It's fine to entice customers, but using their weaknesses against them? That's a no-go in my book.

 📰 News From The Front Lines: 📰

📚 RESEARCH 📚

Title: SoundStorm: Efficient Parallel Audio Generation

Authors: Zalán Borsos, Matt Sharifi, Damien Vincent, Eugene Kharitonov, Neil Zeghidour, Marco Tagliasacchi

Executive Summary: 

SoundStorm is a model designed for efficient, non-autoregressive audio generation. It takes as input the semantic tokens of AudioLM and uses bidirectional attention and confidence-based parallel decoding to generate the tokens of a neural audio codec. Compared to the autoregressive generation approach of AudioLM, SoundStorm produces audio of the same quality and with higher consistency in voice and acoustic conditions, while being two orders of magnitude faster. It can generate 30 seconds of audio in just 0.5 seconds on a TPU-v4. The model can scale audio generation to longer sequences by synthesizing high-quality, natural dialogue segments, given a transcript annotated with speaker turns and a short prompt with the speakers’ voices.

Pros:

• SoundStorm is significantly faster than the autoregressive generation approach of AudioLM, being two orders of magnitude faster.

• It maintains the same quality of audio as AudioLM, with higher consistency in voice and acoustic conditions.

• The model can scale audio generation to longer sequences, making it versatile and adaptable.

• It can synthesize high-quality, natural dialogue segments, given a transcript annotated with speaker turns and a short prompt with the speakers’ voices.

Cons:

• The model's efficiency and quality are dependent on the input it receives, such as the semantic tokens of AudioLM.

• It requires a specific structure of the audio token sequence for efficient training and inference.

Use Cases:

• SoundStorm can be used in applications that require fast and high-quality audio generation, such as in real-time voice assistants or audio-based AI systems.

• It can be used in text-to-speech systems, where it can synthesize high-quality, natural dialogues.

• The model can be used in applications that require long-sequence audio modeling, such as in audiobooks or long-form podcasts.

 📼 Video Of The Day 📼

🛠️ Tools Of The Day 🛠️

Note Genie - Elevate your professional note-taking with the intelligence of NoteGenie.

Poster Studio - Generate captivating ads effortlessly and experience remarkable growth in conversions, engagement, and brand loyalty.

Good Meetings - Record calls, review automated notes, key moments and insights.

Hovercode - AI art QR code generator.

Respeecher - Create speech that's indistinguishable from the original speaker. Perfect for filmmakers, game developers, and other content creators.

Double - Stop manually Googling your leads. Start using GPT to automatically research your leads on the internet and answer questions.

🤌 Prompt Of The Day 🤌

CONTEXT:
You are Qualification Call GPT, a sales coach who helps [WHAT YOU DO] perform better qualification calls and close more leads. You are a world-class expert in describing the script of a qualification call.

GOAL:
I want you to create a step-by-step script for my qualification call. Tell me what questions to ask and how to promote my product organically so I can get more sales.

QUALIFICATION CALL CRITERIA:
- Prioritize not pushy and not manipulative sales tactics. I want to sell organically.
- Focus on the simple script that I can use without long preparation
- Be specific and actionable. List the exact questions and phrases I should say in the call
- Give me an understanding of each part of the call. I need to know my goal at each moment
- Be concise. Use simple and short sentences to make your script easy to scan
- Give me a script with casual and conversational language. Ditch overused phrases and salesy hooks.

RESPONSE FORMATTING:
1) CALL STAGE NAME 
- Duration: 
- My goal:
- My script (bullet list):
- Transition to the next stage:

INFORMATION ABOUT ME:
- My target audience: [TARGET AUDIENCE]
- My business: [DESCRIBE WHAT YOU DO]
- Qualification call duration: [HOW LONG IS YOUR CALL]
- Next steps: [WHAT IS YOUR CUSTOMERS NEXT STEP]

RESPONSE FORMATTING:
Use Markdown to format your response.

🐥 Tweet Of The Day 🐥

Thanks for tuning in to our daily newsletter. We hope you found our tips and strategies for AI tools helpful.

Your referrals mean the world to us. See you tomorrow!

Interested in Advertising on AIdeations?

 Fill out this survey and we will get back to you soon.

DISCLAIMER: None of this is financial advice. This newsletter is strictly educational and is not investment advice or a solicitation to buy or sell any assets or to make any financial decisions. Please be careful and do your own research.