Slide 1

Slide 1 text

Bastian Grimm | Peak Ace AG | @basgr The Rise of AI Strategies & Tips to Drive Growth June 29th 2023

Slide 2

Slide 2 text

A bit of a warmup…

Slide 3

Slide 3 text

Stock Photo or AI? Which image was created by a machine/AI?

Slide 4

Slide 4 text

VS #A #B

Slide 5

Slide 5 text

VS #A #B

Slide 6

Slide 6 text

VS #A #B

Slide 7

Slide 7 text @peakaceag 7 Midjourney V5 has gotten REALLY good… just wow! Our all-time favourite AI text-to-image generation tool that's available right now: Source: &

Slide 8

Slide 8 text @peakaceag 8 Various alternatives, e.g., Dream Studio by Source:

Slide 9

Slide 9 text

Prompting is how humans can talk to AI. It's a way to express what we want and how we want it, usually done with words. Prompt(ing)?

Slide 10

Slide 10 text @peakaceag 10 Need help and/or inspiration? Try DS Prompt Builder: Source: Super-simple drag & drop prompt building based on visual elements to choose style, lightning, etc.

Slide 11

Slide 11 text @peakaceag 11 Creating AI prompts is a key skill for any generative AI Whoever formulates the best, most creative and most unique AI prompts will create the best content (and save AI/API credits) Source: There is a marketplace (, which is dedicated to trading prompts – use it for inspiration!

Slide 12

Slide 12 text

Ok… great. What about real-world use cases though?

Slide 13

Slide 13 text @peakaceag 13 If you need product ads/images fast, try AI-generated product photography using a drag-and-drop approach that allows you to place product imagery in literally any environment you can think of: Source:

Slide 14

Slide 14 text @peakaceag 14 Remove any unwanted object, defect or even people from your picture in seconds Source:

Slide 15

Slide 15 text

Source: Another very practical use-case: Scan me, I am a very beautiful QR Code!

Slide 16

Slide 16 text @peakaceag 16 Recommended use-cases for generative image AI Even though we‘re still at an early stage, the uses mentioned below are very much possible today and allows for time and cost savings, as well as efficiency gains: Header images e.g., for blog posts or newsletters Social images e.g., for your various feeds such as LinkedIn, Insta, etc. Images for slides e.g., for pitch presentations, slide decks or story boards Image modification e.g., clean-ups, object removal, up-scaling, etc.

Slide 17

Slide 17 text @peakaceag 17 Interesting but equally scary use-case from Osaka Univ. “We propose a new method based on a diffusion model (DM) to reconstruct images from human brain activity obtained via fMRI (functional magnetic resonance imaging)” Source: Reconstructing visual experiences from human brain activity offers a unique way to understand how the brain represents the world, and to interpret the connection between computer vision models and our visual system.

Slide 18

Slide 18 text

Speaking of scary stuff…

Slide 19

Slide 19 text

50% of AI researchers believe there’s a 10% or greater chance that humans will go extinct from our inability to control AI. Source:

Slide 20

Slide 20 text

A bit like… catching a flight when 50% of engineers think there's a 10% chance that everyone on it will die. Would you get on the plane? Source:

Slide 21

Slide 21 text

Exaggeration? Maybe…

Slide 22

Slide 22 text @peakaceag 22 Voice- & face cloning is already very much possible Granted, it’s not perfect just yet, but it’ll do the trick. Solutions such as or frameworks such as DeepFaceLab are available for free/at-minimum cost, to everybody: Source: Good afternoon, everybody, this is President Obama speaking! My dear friend Bastian asked me to say hello. De-age the face Replace the head

Slide 23

Slide 23 text @peakaceag 23 AI-powered scam caused $622k fraud In China, AI-powered face-swapping was used to impersonate a friend during a live video call; the scammers received a transfer of 4.3 million Yuan: Source:

Slide 24

Slide 24 text @peakaceag 24 AI-generated image caused a brief stock market dip A fake image showing an explosion near the Pentagon was shared by multiple verified Twitter accounts last Monday, causing confusion and a brief dip in the stock market: Source:

Slide 25

Slide 25 text

So, just imagine what’s out there that we don‘t have access to…

Slide 26

Slide 26 text

Some LLM theory & background knowledge to start with

Slide 27

Slide 27 text @peakaceag 27 What is a large language model (good at)? Source: A LLM is a (deep) learning algorithm that can recognize, summarize, translate, predict and generate text and other content based on knowledge gained from massive datasets.

Slide 28

Slide 28 text @peakaceag 28 This is very important to understand: These LLMs don’t “write” anything. They generate text based on the number of parameters used in their training, using content they've encountered before.

Slide 29

Slide 29 text @peakaceag 29 A timeline of AI and recently released (L)LMs Source:

Slide 30

Slide 30 text @peakaceag 30 Larger data sets = (usually) better results Googles’ MUM uses a text-to-text Transfer Transformer (T5) trained on the Colossal Clean Crawled Corpus (C4), which is significantly larger than BERT's training data Source: (MUM) Basis: Web crawl (220 TB) with approx. 3 billion web pages (BERT) Basis: Wikipedia with approx. 56 million articles VS From a (Google) marketing standpoint it's easy to say that MUM is 10 times smarter than BERT…

Slide 31

Slide 31 text @peakaceag 31 At least equally important: Training data quality “A Comprehensive Analysis of Datasets Used to Train GPT-1, GPT-2, GPT-3, GPT-NeoX- 20B, Megatron-11B, MT-NLG, […]“ Source: There are many more... ▪ Megatron-Turing NLG (Microsoft & Nvidia): 530 billion ▪ Pathways Language Model 2 (Google): 540 billion ▪ WuDao 2.0 (Beijing Academy of AI): 1.5 trillion

Slide 32

Slide 32 text @peakaceag 32 Problems & challenges with using large language models Aside from misinformation, offensive content and hallucination, there’s lots more to know: Source: Computing power The quality of the results improves as the size of the LLMs increases. Gains are incremental, while costs increase exponentially. Biased training data If biased data is used, the system will likely show the same bias when making decisions in practice. Environmental impact Training a large transformer model produces 5,680% more carbon emissions than one human per year. Static perspectives Large language models are usually trained once and used over long periods of time.

Slide 33

Slide 33 text @peakaceag 33 A real-world “bias example” from Amazon The company developed an AI to help scale recruiting – which didn’t work so well… Source: Amazon’s computer models were trained to vet applicants by observing patterns in resumes submitted to the company over a 10-year period. Most came from men, a reflection of male dominance across the tech industry. […] In effect, Amazon’s system taught itself that male candidates were preferable. It penalised resumes that included the word “women’s” and downgraded graduates of two all-women’s colleges.

Slide 34

Slide 34 text

Have I scared you enough to not blindly trust anything AI by now?

Slide 35

Slide 35 text

Now the fun starts… let’s talk tools to drive efficiency & growth!

Slide 36

Slide 36 text @peakaceag 36 Source: – yup… there is! If you prefer searching, this might be of help:

Slide 37

Slide 37 text

Be very, very mindful of the data you put into any of those tools. #confidentiality #dataprotection

Slide 38

Slide 38 text

AI for: Text copy

Slide 39

Slide 39 text @peakaceag 39 Overview of AI-based content generation tools Depending on your needs and desired outcome, I am sure you’ll find one you like:

Slide 40

Slide 40 text

e.g., use Jasper helping you to write sales copy (for outreach) or content drafts for a blog post, etc. What to use it for?

Slide 41

Slide 41 text

Never publish without checking your output first! Expect to check and edit content before it goes live!

Slide 42

Slide 42 text

AI for: Audio

Slide 43

Slide 43 text @peakaceag 43 Create customisable royalty free music uses AI music generation to compose unique mood-based music to suit every part of your creative piece, e.g., video (can be drag and dropped while creating): Source:

Slide 44

Slide 44 text @peakaceag 44 More AI audio & music tools Krisp AI: Removes background voices, noises and echo from all your calls Cleanvoice AI: Removes filler sounds, stuttering and mouth sounds from podcasts or audio recordings Podcastle AI: Studio-quality recording, editing, and seamless exporting – in a single web platform Royalty-free music, AI generated based on mood, genre and length. To write, record, transcribe and edit video and podcasts Soundbite AI: Turn new and existing audio and video content into ready-to- edit social media posts, and summaries.

Slide 45

Slide 45 text

AI for: Video

Slide 46

Slide 46 text @peakaceag 46 Want to boost personalised interactions with customers? Maverick records once but automatically personalises for all, enabling videos at scale. Source:

Slide 47

Slide 47 text @peakaceag 47 The Peak Ace team used for a cost-effective localisation of video ads assets: Step 1: Transcribe original script and localise into desired language Step 3: Input your localised copy and export script Step 2: Decide on the voice and style you'd like to use Step 4: Add the new script to your video and adjust until it sounds perfect New FR version Our task/challenge: A client wanted to use a video in different regions but didn’t have local employees to voice the scripts. Original DE

Slide 48

Slide 48 text @peakaceag 48 Other video AI solutions worth checking out:

Slide 49

Slide 49 text @peakaceag 49 Gen-2 by Runway in particular looks crazy cool: Just look at how it transfers the style of any image or prompt to every frame of a video: Source: Source Video Generated Video Driving Image

Slide 50

Slide 50 text

AI for: (More) Productivity

Slide 51

Slide 51 text @peakaceag 51 Compose AI: A Chrome extension to make you fast(er) A free Chrome extension to automate your writing Source:

Slide 52

Slide 52 text @peakaceag 52 MailMaestro: Bringing OpenAI LLMs directly to Outlook Generate custom emails and responses based on short prompts; create one click email thread summaries including action points for the receiver: Source:

Slide 53

Slide 53 text @peakaceag 53 Create more, better LinkedIn content with AI using Taplio Establishing a personal brand on LinkedIn is a key pillar of growth. Stop wasting endless hours writing your next post and get fresh ideas using Taplio’s content inspiration layer: Source:

Slide 54

Slide 54 text @peakaceag 54 ExcelFormulaBot: turn your problem into a formula Transform your text instructions into Excel formulas in seconds (yep, I know: ChatGPT can do that as well…) Source:

Slide 55

Slide 55 text @peakaceag 55 for automated meeting notes & summaries Your meeting assistant that records audio, writes notes, automatically captures slides, and generates summaries: Source: Word of advice (and caution, again): It's super practical but getting consent to record is important – especially since some services can automatically send meeting summaries by email afterwards! Also, I am not a lawyer.

Slide 56

Slide 56 text

AI for: No Code

Slide 57

Slide 57 text @peakaceag 57 Grow faster with No-Code ML Create custom ML models in minutes – no technical background needed. Just connect with a data source of your choice (e.g., CSV, SQL, BigQ) and you’re go to go: Source:

Slide 58

Slide 58 text @peakaceag 58 Things Peak Ace has done with Akkio and their models: 1. Sales forecasting – e.g., understanding when customers are likely to buy. 2. Churn prediction – e.g., finding out which customers are likely to cancel next. 3. Lead scoring – e.g., understanding which sources generate the highest purchase intent.

Slide 59

Slide 59 text

AI and: ChatGPT

Slide 60

Slide 60 text @peakaceag 60 ChatGPT is currently using OpenAI’s GPT-4 model Including capabilities for browsing (Bing) as well as plug-in usage Source:

Slide 61

Slide 61 text @peakaceag 61 What makes a good prompt and how do you structure it? Some tips to create better and more efficient prompts for ChatGPT (h/t Mike King): Source: Role Who is ChatGPT creating as? Context What is the situation it is creating for? Instructions What specifically do you want it to do? Format How do you want it to return its response? Examples Samples of the output that you expect. Constraints What should ChatGPT not do?

Slide 62

Slide 62 text @peakaceag 62 ChatGPT combined with AIPRM (prompt mgmt. plug-in) Access a curated selection of ChatGPT prompts specifically designed for SEOs, marketers, sales, support, copywriting and more : Source:

Slide 63

Slide 63 text @peakaceag 63 ChatGPT browsing mode using Bing search: Essentially, OpenAI uses Bing search to enrich output based on automatically generated queries taken from the original ChatGPT prompt:

Slide 64

Slide 64 text @peakaceag 64 ChatGPT plug-ins for more productivity/efficiency e.g., LinkReader allows direct interaction with PDFs, Word documents, etc.

Slide 65

Slide 65 text

Direct integrations into LLMs through plug-ins are going to significantly change the web’s ecosystem

Slide 66

Slide 66 text

AI vs. Search

Slide 67

Slide 67 text @peakaceag 67 The "new Bing" – Microsoft going all-in on AI in search Bings’ integration of OpenAI’s LLM into its search engine kicked-off a whole series of changes for search in general (some changes are still to come):

Slide 68

Slide 68 text @peakaceag 68 It’s still early days, but initial ad integrations are available How will targeting work? What’s the model? How about bidding? So many questions … Source:

Slide 69

Slide 69 text @peakaceag 69 Naturally, Google has something brewing as well… Meet Google Bard (which is not officially available in Europe just yet…) Source: Just use a VPN service of your liking and set your IP origin to the US, or Switzerland even.

Slide 70

Slide 70 text @peakaceag 70 Google‘s Search Generative Experience (SGE) Join the waitlist (if you’re not already on it): (US IP/VPN needed) Source:

Slide 71

Slide 71 text @peakaceag 71

Slide 72

Slide 72 text @peakaceag 72

Slide 73

Slide 73 text @peakaceag 73

Slide 74

Slide 74 text

Your marketing team needs to be thinking about adapting their SEO strategy, as obtaining organic traffic is likely to get much more difficult in the future. Why should you care?

Slide 75

Slide 75 text

Organic search results will change, but so will ad creation and the integration of ads into the search results. Google to introduce generative AI for ads as well

Slide 76

Slide 76 text @peakaceag 76 Meanwhile… automated RSAs based on OpenAI’s API Leverage OpenAI’s API to use the maximum number of Google responsive search ads assets and, in turn, boost your paid search campaigns: Source: & Another interesting idea:

Slide 77

Slide 77 text

Built from the ground up with true multimodality, high efficiency in using external tools/APIs, memory capabilities, etc. Google also is working on a brand-new model: Gemini

Slide 78

Slide 78 text

A multimodal AI can ingest knowledge/input from multiple sources (e.g., text, audio, video, gesture, …) and utilise it to solve tasks involving any modality Multimodality?

Slide 79

Slide 79 text

So, where are AI & LLMs headed in the future?

Slide 80

Slide 80 text @peakaceag 80 Amazon entering the game with impressive initial results And their very own (pretty horrible) on-site search will most likely undergo a full makeover as well: Source: & Amazon has released a new LM that outperforms GPT-3.5 on ScienceQA by 16 percentage points, or with 75.1% to 91.6% accuracy

Slide 81

Slide 81 text @peakaceag 81 Microsoft to broadly roll-out their MS365 Copilot This isn't strictly a topic for growth, but "interacting with AI“ will soon be part of your daily (work) activities – and you might not even know it: Sources: & Invoking Windows Copilot is easy – the button is front and center on your taskbar – simple to find and use. Once open, the Windows Copilot side bar stays consistent across your apps, programs and windows, always available to act as your personal assistant.

Slide 82

Slide 82 text

With one significant goal in mind: improving themselves Models will generate their own training data

Slide 83

Slide 83 text

Because today’s LLMs make stuff up Models will fact-check themselves

Slide 84

Slide 84 text

Shifting towards models which only use the most relevant subset of their parameters vs. running through every single one > Massive sparse expert models New LLM architecture?

Slide 85

Slide 85 text @peakaceag 85 Planning for Artificial General Intelligence and beyond “As our systems get closer to AGI, we are becoming increasingly cautious with the creation and deployment of our models.” Source: The first AGI will be just a point along the continuum of intelligence. We think it’s likely that progress will continue from there […] If this is true, the world could become extremely different from how it is today, and the risks could be extraordinary. A misaligned super intelligent AGI could cause grievous harm to the world; an autocratic regime with a decisive superintelligence lead could do that too.

Slide 86

Slide 86 text @peakaceag 86 The best part of this OpenAI post though? Source: Successfully transitioning to a world with superintelligence is perhaps the most important— and hopeful, and scary—project in human history. Success is far from guaranteed, and the stakes (boundless downside and boundless upside) will hopefully unite all of us.

Slide 87

Slide 87 text


Slide 88

Slide 88 text @peakaceag 88 Meanwhile: World’s first AI-flown fighter jet can dogfight In a joint project between DARPA and the US Air Force, a special aircraft called the X-62 Vista became the first tactical aircraft to be piloted by AI: Source: The jet was under the control of one of four different AI algorithms at any given time during the tests. […] This included dogfighting in simulated combat missions, as well as takeoffs and landings

Slide 89

Slide 89 text

The Online Marketer’s AI Playbook: Thank you!