AI Swarms: Are They the Future of Business & Marketing ?

Welcome back to your favorite AI marketing newsletter! This week, we're buzzing about the potential of collaborative AI swarms and wondering what would happen if every company decided to automate with AI. Let's dive in!

Last week, we learned that agents shine when they are specialized. But imagine what would happen if they paired up in collaborative swarms that specialize in different areas. The result could be more valuable than a single, general AI that can do everything.

a ChatGPT system could select the best models for the job and iterate new prompts across dozens of AI models and tools. The possibilities are beyond exponential, which means the future of AI is definitely not niche, but rather a swarm of collaborative, specialized agents.

Of course, not every organization is ready for this level of AI implementation. But we believe that the solution lies in building confidence through an approach that allows the agent to momentarily break out of its process and ask a human for assistance before reverting back to automation. It's like taking a refreshing break from the current and pervasive AI agents implementation scenario.

One of the biggest challenges facing businesses today is keeping up with the pace of technological advancements. But with AI swarms, businesses can implement AI more effectively and efficiently, by trusting and self-selecting the best models for the task at hand. This streamlines the process, allowing businesses to quickly adapt to changing market demands and gain a competitive edge.

Now, for the trillion-dollar question: if all companies were to use AI swarms to automate their workflow, how would they differentiate themselves and compete? What about jobs?

Some say companies could differentiate on innovation and customization, as well as integration and adaptability. But we're not convinced. After all, if AI swarms can innovate on their own, then the ultimate impact on differentiation will still depend on how companies manage and direct these AI systems.

So, here's what I think will differentiate companies using AI swarms:

• Strategic Direction: Businesses that can effectively align their AI swarms with their strategic goals will differentiate themselves from competitors.

• Human Expertise and Creativity: Companies that can effectively harness the strengths of both AI and human employees will have a competitive advantage.

• Proprietary Data: Access to unique or proprietary data sources will allow businesses to leverage their AI swarms more effectively.

• Collaboration and Partnerships: Strong partnerships and collaborations can provide businesses access to new technologies, data sources, or markets.

• Agility: Companies that can rapidly execute on their AI-driven innovations will be better positioned to compete.

• Data Quality and Security: Maintaining robust data security practices and high-quality data will be important to customers.

• Ethical AI Practices: Public scrutiny of ethical implications means that businesses using ethical AI practices will be more attractive to customers.

Even if every company uses the same AI swarm technology, the way they implement it and the specific tasks they use it for will still differ. And, there will always be room for companies to differentiate themselves based on other factors like branding, customer experience, and company culture. So while AI swarms may streamline certain aspects of a company's operations, they cannot replace the human touch that many consumers still value.

In terms of job loss and creation, the widespread adoption of AI swarms is likely to lead to a shift in the labor market.

Some jobs may become obsolete, particularly those involving repetitive tasks or simple decision-making.

However, new job opportunities will also emerge, such as

• AI Agent Managers, who are professionals that work on teaching AI systems how to perform tasks, understand context, and improve their performance;

• AI ethicists and policy advisors, who are experts that ensure that AI systems operate ethically, comply with regulations, and address societal concerns;

• AI integration & quality specialists, who are individuals that help businesses integrate AI into their existing processes, ensuring smooth adaptation and maximizing efficiency; and

• Human-AI collaboration managers, who are professionals that facilitate effective collaboration between human employees and AI systems, ensuring that the combined workforce achieves its full potential.

Additionally, there will always be a need for human oversight and decision-making, so jobs that require critical thinking will likely remain in demand.

To forecast the impact of AI swarms on organizations and jobs, let's use ChatGPT to analyze the short, medium, and longer-term implications.

Specifically in the marketing world, AI Swarms could have the potential to automate most of these tasks:

Source

I used "could" because we still have a long way to go. The founder of BabyAGI acknowledges the slow but steady progress being made to extract value from his Agent. He tweeted, "Success rate on completing full task list is still really low (~10%), but starting to do some pretty cool stuff! This one successfully did a web search, then scraped the site and summarized the results. (It unexpectedly restarted)." (Source)

AI Agents will have access to various tools to achieve their tasks, including audio and voice. AI voice cloning is on the rise, from scams to music. (Source)

I have spent the week testing various audio synthesis models and was stunned by the potential. For example, I built a dragon's den equivalent during the week to analyze and give feedback on business and product ideas, with 2PAC as a critic.

This is not new (considering it is only two months old 🤣). You could use ChatGPT to analyze your idea, but what made it magical was hearing 2PAC voicing his concerns and needs. (Listen here )

Suddenly, an emotional connection was made. This is something you don't have with ChatGPT.

And I'm just scratching the surface here, as I did not convey emotions in my prompt or ask for emotions in return. I tried the experience three times, and the replies from 2PAC became kinder and kinder for some reason. While this is not statistically significant, I am going to explore this further.

When Agents start accessing audio models, we will reach a completely different level of user experience and engagement. Some artists have piggybacked on this in a smart way. Grimes, for example, said on Monday that anyone could use her AI vocals. By Wednesday, musicians in Web3 had started dropping Grimes AI tracks. We are at the cutting edge. If you want to try it yourself, here's the blueprint

These developments are prompting governments to set up task forces. The US Senate has passed a bill to form an AI task force (Source) to specify how best to reduce hazards to privacy, civil freedoms, and due process or to strategize their emergency five-year plans (Japan).

On Thursday, the EU released a new draft of legislation that identifies copyright protection as a core piece of the effort to keep AI in check.

Companies deploying generative AI tools such as ChatGPT will have to disclose any copyrighted material used to develop their systems, according to an early EU agreement that could pave the way for the world's first comprehensive laws governing the technology.

AI tools will be classified according to their perceived risk level, from minimal to limited, high, and unacceptable.

Areas of concern could include biometric surveillance, spreading misinformation, or discriminatory language. While high-risk tools will not be banned, those using them will need to be highly transparent in their operations (source)

Hopefully, the development of AI systems will be supported by the best minds on the planet, such as computer scientist Yejin Choi. She aims to demystify the current state of massive artificial intelligence systems like ChatGPT and has highlighted three key problems with cutting-edge large language models. In her TED talk (Link), she welcomes us into a new era where AI is becoming almost like a new intellectual species, and she identifies the benefits of building smaller AI systems trained on human norms and values.

Surprisingly, even small businesses are embracing AI, according to a survey conducted by Clutch. Almost all small business leaders (82%) anticipate that AI will disrupt their operations in the next few years, and many think that AI will completely transform their businesses during that time. Business leaders shared that they were excited, optimistic, and motivated about AI, while less than 10% of respondents said they felt overwhelmed, frightened, or worried about generative AI technology. Small businesses believe that AI is best suited for responding to customers with chatbots, creating business presentations, and writing marketing collateral.

Source

To help you stay on the cutting edge of AI, I offer "Briefing" sessions for teams of 3 to 20 people. During these sessions, which last 90 minutes and take place on Zoom, I cover the following topics:

• AI integration in the marketing workflow

• ChatGPT and its plugins

• AI agents and infrastructure

• AI for audience analysis

• AI for growth with predictive modelling

If you're interested in booking a session, please reach out here: https://calendly.com/getinference.

Behold these top nine AI wonders of the week that are sure to make you go "Ooh!" and "Aah!":

TOOLS

📌LeiaPix Converter: Turn your favorite images into stunning depth animations. (Link)

📌MyVocal AI: Use AI to clone your voice for singing or content creation within 60 seconds. (Link)

📌Stable Art: Create easy stable diffusion and community models with 100 free images per month. (Link)

📌DreamboothR: Train your own Dreambooth models. (Link)

📌Podsmart: Use OpenAI Whisper for state-of-the-art audio transcription and transcribe a one-hour podcast in just two minutes. (Link)

📌Bark: This universal text-to-audio model can generate highly realistic, multilingual speech as well as other audio, including music, background noise, and simple sound effects. (Link) Note that this demo should only be used for research purposes.

📌SiteGPT: Use ChatGPT for every website. (Link)

📌Evoto AI: Polish your photos to perfection with AI. (Link)

📌InVideo AI: Create publish-worthy videos on day one. (Link)

So, there you have it - nine AI-powered treats to make your life easier, more efficient, and dare we say, a little more entertaining! 🎉

SPOTLIGHT

In this week's spotlight, we have (again) several exciting developments in the world of AI:

🔦Meta introduces Codex Avatars, which are lifelike avatars that can be created using a smartphone.

🔦Italy welcomes back ChatGPT.

🔦A team of researchers generates surprisingly great 3D models out of text prompts, indicating a promising future for this technology.

🔦ElevenLabs AI can now clone the voice of someone who speaks English and have them say things in a language they don't speak.

🔦Following the viral success of an AI-generated Drake song, a new website called AIhits.co has been launched to help people discover viral music generated by AI.

🔦Track Anything is a tool that combines Segment Anything with video object tracking and segmentation, making it great for video editing.

🔦Stability AI releases DeepFloyd IF, a powerful text-to-image model that can smartly integrate text into images. (Link )

AGENTS

👉 This is mega: Large Language Models (LLMs) are very impressive, but they can be made even more powerful if we give them skills to accomplish specialized tasks. The gradio_tools library can turn any Gradio application into a tool that an agent can use to complete its task. For example, an LLM could use a Gradio tool to transcribe a voice recording it finds online and then summarize it for you. Or it could use a different Gradio tool to apply OCR to a document on your Google Drive and then answer questions about it. (Link)

👉 If this is not enough, you can create custom tools for chatbots in LangChain (Link)

👉 A swarm of agents is coming. Here is the research paper (Link)

👉 Lars is building another framework to make it easy to create specialized agents , it's still in its early stages (Link)

👉 Vector search is good for capturing semantically similar texts, but often queries specify desired attributes like time, authorship, or other "metadata" fields, which vector search is not great at. Enter SelfQueryRetriever from Langchain again. What geniuses!

👉 Join our AI agents group on LinkedIn for daily deep dives on these subjects (Link)

AMAZING AI

👉 Kaiber is great for performers/music videos (Link)

👉 Amazing animation on LinkedIn (Link)

👉 Curious about what the world is building with generative AI? Here's a Twitter thread worth a look (Link)

👉 Pushed Runway #gen2 to make a movie trailer. #aicinema is finally here! Every shot made from text prompts, except one iconic shot you all know, done with #gen1. Made possible by @runwayml (Link)

👉 AI video has started to produce mind-blowing results and could eventually disrupt Hollywood. It seems like a toy right now, but things are changing fast. Here are the best AI videos I've found (Link)

👉 A 23-second glimpse into the future: @pwang_szn built FigmaChain and it's incredible. Here's how it works: 1. Import Figma design with @LangChainAI loader 2. Chat to make design changes via @streamlit 3. Updates design + returns code (Link)

👉 Top 4 Image to Video AI Tools: Create AI Animation for FREE (Link)

👉 This AI Voice Generator is Emotional (Link)

DISCUSS

👉 Check out this list of 11 interesting new careers that could emerge in the AI revolution: (Link)

👉 Why is everyone so excited about Llama and other non-ChatGPT LLMs? Find out here: (Link)

👉 Looking for the best current Local LLM to run? Check out this discussion: (Link)

👉 The new LLM champion is StableVicuna, with an impressive 13 billion parameters! (Link)

LEARN

👉 Discover how to reach every customer with generative AI through segmentation experiments (Link)

👉 Introducing Scrapeghost, a new library for scraping websites using #GPT4! This library can scrape structured data from HTML without writing page-specific code (Link)

👉 Learn about ChatGPT explained simply (Link)

👉 Take the ChatGPT Prompt Engineering for Developers short course (Link)

👉 Utilize GPT-4 as a compression mechanism to significantly expand its effective context size by requesting it to compress your prompts using its own abbreviations. Check out this example where GPT-4 compressed a 3,500 character background text into just 395 characters using its own language! (Link)

👉 Check out some observations and experiments on the ChatGPT Default mode, now including Browsing mode enabled for plus users (Link)

👉 Get the Midjourney Prompt collection v5 for more prompts (Link)

👉 Create realistic avatar videos with ChatGPT using the AI Video Generator (Link)

WRAPPING UP

You made it to the end of GetInference's AI for Marketing newsletter! That's more than we can say for some of our AI experiments... but we won't talk about those.

If you found this newsletter helpful, please share it with your friends and colleagues. They'll thank you, and we'll thank you even more if they subscribe (Link) .

Don't forget to join the AI Radar (Link) and GetInference Academy (Link) .

Thanks for your support! We couldn't do this without you (and ChatGPT, really ? nah, human made). ❤️