AI news for marketers, communicators and content creators from Silverbeard AI

February was a month of significant developments in the generative AI world, with advances in a number of areas - AI generated video, the amount of information that can be included in a prompt and speed being the most important. Here's a summary of the key trends and their significance:

1. The competition heats up as Google gets going: Gemini Advanced and Gemini Pro 1.5 launched
2. AI video generation takes centre stage: Sora demands a starring role
3. The not so secret agents are coming: OpenAI, Microsoft and Hyperwrite are on the case
4. AI and job displacement: Klarna’s AI bot does the work of 700 people
5. There may be trouble ahead: Google AI images forgo accuracy for diversity
6. Speed and memory: Groq, Google and OpenAI build for tomorrow
7. Content is king: Google in deal with Reddit
8. Ethics and fraud prevention: Meta working on identifying AI content
9. Apple is just warming up
10. Collaborations and partnerships summary: Jasper/Clipdrop, Perplexity/ElevenLabs, Midjourney/X, Microsoft/Mistral Large

1. The competition heats up as 谷歌 gets going - Gemini Advanced and Gemini Pro 1.5 launched:

• What happened: Google finally launched its answer to OpenAI ’s ChatGPT and 微软 Copilot, rebranding Bard as Gemini and releasing “GPT 4 killer” Gemini Advanced and Gemini Pro 1.5. Confusing? Yes, we think they could do with some brand advice too...
• Why it’s important: Having invented much of the tech that made generative AI possible, Google has been noticeably late to the party, allowing Open AI to lead the way for over a year. Google has been grappling with a number of issues, not least the threat AI poses to their search revenue, which was reported to be $44 billion for the quarter to September 2024 - almost 60% of their total revenue. But, this launch signals they’re not ready to go the way of Kodak in the 1990s. And, for workers and teams, Gemini Advanced provides a decent alternative to Open AI’s GPT4 version of Chat GPT and Microsoft’s Copilot - especially if you’re a user of Google Cloud and their suite of productivity apps.
• Now, if all those bright minds could apply themselves to their naming strategies, so the rest of us could more easily distinguish Gemini services from Gemini models…
• Read more about Gemini Advanced
• Read more about Gemini 1.5

2. AI video generation takes centre stage - Sora demands a starring role

• What happened: OpenAI unveiled Sora, an impressive (but not yet available to the public) AI video generator capable of creating short videos from text descriptions.
• 阿里巴巴集团 also announced their entry with EMO, an AI video generator that syncs video talking heads and voice with a fair amount of realism.
• Why it’s important: Many of us had 2024 down as “the year of AI video”. Sora, although not yet released, may well be that confirmation. Certainly the test videos that flooded X/Twitter were enough to convince Tyler Perry, actor and owner of Tyler Perry Studios , to put an $800m, 12 sound stage studio expansion, in planning for the last 4 years, on indefinite hold.
• A world where high quality video production is just a natural language prompt away appears to be within reach - although it’s not here quite yet. But, when it is it will have a profound effect on video production studios (for good and for bad), as well as marketers and content creators whose only limits will be their understanding of their audience and their imagination…
• Read more about Sora
• Read more about Tyler Perry Studios reaction to Sora
• Read more about Alibaba's EMO

3. The not so secret agents are coming - OpenAI, Microsoft and Hyperwrite are on the case

• What happened: OpenAI announced that AI agents or assistants - where AI is used to complete a task rather than just suggest a solution - is an area of importance and they are focused on automating tasks on your computer or mobile, as well as on the web.
• Microsoft rebranded Bing as Copilot and positioned it as “Your everyday AI companion” in their Super Bowl ad.
• HyperWrite (OthersideAI) introduced AI assistants as part of their Agent Studio platform.
• Why it’s important: Perhaps more than gloriously generated content, AI agents or assistants that can genuinely shift the admin load off our desks and out of our lives will be the real game changer. This is AI that does rather than AI that just says. The shift from AI that can suggest flights and hotels in response to a travel itinerary request to AI that actually books and pays for the flights and hotels, at the best price, requesting your preferred seat/room etc.
• The application of AI assistants will be ubiquitous. At work we might use it to onboard new joiners, collect together and make sense of disparate sources of data before writing a board report, automatically collect our expense receipts and bookings to compile and submit an expense claim. In fact any task that could be intelligently automated. Leaving us with more time for strategic, creative or innovative thinking. Well, that’s the idea...
• Read more about OpenAi and AI Assistants
• Read more about Microsoft Copilot

4. AI and job displacement - Klarna's AI bot does the work of 700 people

• What happened: Klarna reported that their new Open AI-powered bot could replaced the work of 700 people, after just a month of use, sparking further debates about AI's impact on the workforce.
• Why it’s important: While some were sceptical of Klarna’s conclusions and timing - Klarna was reporting on only the AI bot’s first month of use - few would doubt that a revolution in work and roles is on its way. It will deliver enormous cost and productivity benefits alongside an impact on jobs that will require collaborations between companies and the wider society to address. No doubt this latter need will be lead by companies that place a higher value on their brand reputation.
• Read more about Klarna's AI bot

5. There may be trouble ahead - Google AI images forgo accuracy for diversity:

• What happened: Google Gemini AI generated images were suddenly found to prioritise diversity over historical accuracy, for example generating images of soldiers in Nazi uniform from a diverse range of ethnic backgrounds, causing a 4.7% stock drop, while the Nasdaq rose 1.7%.
• Additionally, there were incidents involving ChatGPT suddenly speaking nonsense and talking “Spanglish” as well as Microsoft with the reemergence of the abusive “Sydney" personality from Bing.
• Why it’s important: The incidents, especially the Gemini images issue, raise ethical questions about who sets and who should set the ethical parameters for AI, as well the big question: are we really in charge at all? Meanwhile, Google Gemini’s ability to generate images of people is still offline…
• Read more about the Gemini images issue
• Read more about the OpenAI 'Spanglish' issue
• Read more about the Microsoft 'Sydney' issue

6. Speed and memory - Groq, Google and OpenAI build for tomorrow:

• What happened: Groq , which designs and manufactures computer processors specifically for AI language models, announced a significant speed increase through their new processor, while Google (with Gemini 1.5) and OpenAI (with their new "memory" feature) made strides in extending AI's prompting capabilities.
• While most AI runs on “GPU” processors, Groq (that’s Groq with a “Q”, not Musk’s newer company with a “K”) created what it termed LPUs, or “language processing units” which, unlike GPU processors, are designed specifically for running generative language AI and reportedly would run ChatGPT 13 times faster. In practice, that means it “answers with hundreds of words in less than a second”.
• In addition, Google’s Gemini 1.5 model reportedly comes with a market-leading “context window of up to one million tokens”. This is a measure of how much information you can put in to your prompt for a question. One million tokens is enormous - the equivalent of around 750,000 words, which means you could comfortably include "War and Peace" by Leo Tolstoy or the complete? "Lord of the Rings" trilogy. For comparison, the context window for Chat GPT Plus is 32,000.
• Separately, OpenAI announced they would slowly roll out a new ‘memory’ feature that enables users to save information from one prompt session for use in further prompt sessions.
• Why it’s important: These advancements pave the way for more capable and practical AI applications, particularly in the realm of AI assistants. For example, for an AI assistant to be useful, it has to be able to respond without a time lag and ‘remember’ critical information, whether that’s your payment details or how you like your eggs cooked! For marketers and content creators, AI that is able to ‘memorise’ extensive documentation or videos, to ensure that a brand is presented consistently and clearly across many different media and formats, would be a godsend!
• Read more about Groq

7. Content is king - Google in deal with Reddit:

• What happened: Google acquired Reddit, Inc. content in a deal reportedly worth $60 million a year.
• Separately, Google is also developing an artificial intelligence tool for news publishers that can generate article text and headlines.
• Why it’s important: There are three sides to the generative AI triangle: processors, models and data. Google has long held an advantage in the latter. Google stills dominates search and has scraped nearly the entire public internet for over 25 years to do so. And, it also owns YouTube which hosts a billion videos, complete with viewing habits and a rich source of an increasingly valuable format for AI model training. But,?perhaps mindful of the copyright traps others are falling in (such as OpenAI Vs 纽约时报 ) they’re not resting on their laurels.
• Read more about Google's deal for Reddit content

8. Ethics and fraud prevention - Meta working on identifying AI content:

• What happened: Meta announced it is working with other industry partners on common technical standards to identify AI-generated content, to combat ‘deepfakes’.
• 20 leading technology companies announced the “Tech Accord to Combat Deceptive Use of AI in 2024 Elections” - a set of commitments to deploy technology countering harmful AI-generated content intended to deceive voters.
• More than 300 experts, including US politician Andrew Yang, MIT lab computer scientist Joy Buolamwini, British computer scientist Stuart Russell and psycholinguist Steven Pinker, released an open letter urging immediate action from governments to combat deepfake content.
• Why it’s important: All three initiatives come in the wake of a surge in harmful deepfakes involving sexual imagery, fraud, and political disinformation in a year when the stakes are even higher. There’s not just a US Presidential Election, but more than 4 billion people - 40% of the world's population - are eligible to vote globally. Building an AI infrastructure that will instil trust in AI - that includes clarity as to what is 'made by AI' and what isn’t - is essential to the success and widespread use of AI.
• Read more about Meta and identifying AI content
• Read more about Microsoft and the Tech Accord to Combat Deceptive Use of AI
• Read more about the open letter from 300 experts on combating deep fakes

9. Apple is just warming up:

• What happened: Though known for prioritising quality over speed, 苹果 hinted at major AI announcements at their Worldwide Developers Conference 24 in June.
• They also released an AI image editing tool called MGIE in research mode and reportedly refocused their efforts on AI, shutting down their autonomous car project.
• Why it’s important: Fear not Apple fan girls and boys, these signs suggest Apple will introduce significant AI advancements this year, including Siri enhancements, an AI system for generating videos and images, and deeper AI integration into their operating systems and applications.
• Read more about Apple's MGIE
• Read more about Apple's plans for AI

10. Collaborations and partnerships summary - Jasper/Clipdrop, Perplexity/ElevenLabs, Midjourney/X, Microsoft/Mistral Large:

• What happened: February saw reports of a number of notable or potential integrations, collaborations and mergers, including Jasper and Clipdrop, Perplexity and ElevenLabs, MidJourney and X, and Microsoft and Mistral Large.
• Jasper , the AI word generator used by many marketers and content creators acquired ClipDrop , an AI image creation and editing platform, from Stability AI, that will make it even more useful to that user base. Perplexity and ElevenLabs partnered up to create a daily podcast called “Discover Daily”. Musk, owner of X /Twitter said they were in talks with AI Snap , one of the leading AI image generation tools, and Microsoft broadened the AI models it offers its customers with a deal to provide Mistral AI -Large on Microsoft Azure , Microsoft’s cloud platform.
• Why it’s important: The speed of progress in AI is high, as is the cost of investment. We will see more collaborations and acquisitions as companies providing AI services seek to increase their reach and engagement by hitching a ride from better distribution networks and creating services that are tailored to the needs of their customers.
• Read more about Jasper AI's acquisition of Clipdrop
• Read more about Perplexity's collaboration with ElevenLabs
• Read more about a potential link up between MidJourney and X/Twitter
• Read more about the deal between Microsoft and Mistral

That’s it for this month. I hope you found it interesting and helpful. While you’re here, why not follow Silverbeard AI for regular updates on generative AI, especially those that have a practical effect on marketing, content creation and running a business.

You may also be invested in courses we’re running for marketers and content creators in partnership with Emarketeers . Details are available here:

https://www.emarketeers.com/training-courses/ai-for-marketers

https://www.emarketeers.com/training-courses/ai-for-content-creators-copywriters