The captain's log
Jeremy Rosenberg
Generative AI, Marketing, Brand & Communications Consultant, Trainer & Coach
AI Weekly 23.10.24
A bracing sail through the seas of generative AI, handily mapped out under the following headings:
- AI and Content Creation
- AI and Marketing
- AI Models and Tools
- AI Regulation and Ethics
- AI Adoption
- AI and Agents
- AI, Health and Education
AI and Content Creation
Runway , known for its leading gen AI video models, introduced Act-One, a new state-of-the-art tool for generating expressive character performances in animations using video and voice inputs, including of live performances. It will be available inside Gen-3 Alpha.
Adobe teased AI tools that build 3D scenes, animate text, and make distractions disappear.
Adobe expanded its Firefly family of generative AI models with a new video generation model aimed at enterprises concerned about copyright and other problems associated with generative AI technology.
Stability AI announced Stable Diffusion 3.5, an updated version of their open source gen AI image models. Stability AI says the models excel at customisation, efficiency, diversity and creating a wide variety of styles. The models are freely available for both commercial and non-commercial use in most cases.
Midjourney is planning to release an upgraded web tool that’ll let users edit any uploaded images from the web using Midjourney’s generative AI.
Meta is the latest Gen AI model maker to partner with a Hollywood production company to help develop its video model. In this case Meta has announced it is working with Blumhouse Productions to test MoviGen.
AI and Marketing
亚马逊 Ads announced AI creative studio and Audio generator—two new generative AI tools to help advertisers create a more dynamic and engaging experience for customers.
How GenAI is transforming the structure, roles and skills of marketing teams.
How AI and influencer marketing are transforming audio into the most effective ad medium.
AI Models and Tools
Anthropic introduces a developer beta version of ‘computer use’ - an agent that can perform tasks with your computer files and the web - as well as new versions of its highly rated Claude 3.5 Sonnet and Claude 3.5 Haiku gen AI models.
Chat GPT’s newest feature - Canvas - has been updated, and can now show tracked changes
Meta was heavily criticised for calling its Llama models, “open-source”, reports the Financial Times .
谷歌 ’s NotebookLM - now no longer labelled “experimental” and used in over 800,000 organisations, according to Google - gets a big update.? For example, audio overviews - the feature that turns your source material into a podcast-style discussion between two people - are now customisable. You can now set the expertise level and steer the focus of the conversation, unlocking a number of new business uses. You can also continue to listen to the audio overviews, while doing something else within NotebooksLM. A premium version for business, unsurprisingly called NotebookLM Business, is coming soon.
Perplexity adds internal knowledge search and Spaces.
谷歌 reshuffled its Gemini team into Google DeepMind Division, as it continues to transition from an “AI research labs” to an “AI product company”. The restructure will reportedly improve feedback loops, enable fast deployment of new models in the Gemini app, make post-training work proceed more efficiently.
OpenAI released a ChatGPT app for Windows, but only for paying customers and even then, with some features missing
All Chat GPT Plus users in the European Union , Switzerland, Iceland, Norway and Liechtenstein now finally have access to Advanced Voice Mode.
微软 ’s “best bromance in tech” with OpenAI seems to be frayed as OpenAI has tried to change its partnership deal with Microsoft and the software maker has tried to hedge its bet on the AI start-up, reports the NYT in an exclusive.
苹果 is due to make Apple Intelligence available to the public in a software update next week (if you have a sufficiently recent iPhones, iPads and Macs), but don’t mismanage your expectations. Bloomberg News ’s Mark Gurman, writes that internal Apple research found that ChatGPT is 25 per cent more accurate than Siri will be with Apple Intelligence and can answer 30 per cent more questions - in fact, some at Apple believe they are two years behind the industry leaders.
Longer term, of course, it would be unwise to bet against Apple’s reach and resources…
AI Regulation and Ethics
Anthropic , the maker of the highly praised Claude LLMs, revised its safety guidelines to manage what it believes are escalating AI risks.
Elon Musk’s X announced an update to its privacy policy, permitting third-party companies to access users' public posts for AI model training.
Penguin Random House amended its copyright rules to protect authors from AI.
Dow Jones & Co., publisher of the The Wall Street Journal and the New York Post , sues AI search startup Perplexity for copyright infringement.
Abba’s Bj?rn Ulvaeus, the actor Julianne Moore and the Radiohead singer Thom Yorke are among 10,500 signatories of a statement from the creative industries warning artificial intelligence companies that unlicensed use of their work is a “major, unjust threat” to artists’ livelihoods.
The Financial Times reports that the UK Government is set to consult on a scheme that would allow AI companies to scrape content from publishers and artists unless they “opt out”, in a move that would anger the creative industry.?
AI Adoption
The The Brookings Institution research on “Generative AI, the American Worker and the Future of Work” reported that existing generative AI technology already has the potential to significantly disrupt a wide range of jobs and concludes most companies are not prepared for the disruption. They believe that more than 30% of all workers could see at least 50% of their occupation’s tasks disrupted by generative AI.?
Harvard Business Review article: How to Use AI to Build Your Company’s Collective Intelligence.
Two recent surveys - from Salesforce and Theta Lake - highlight the need for consistent, high-quality data in building reliable AI systems.
JPMorgan Chase Commercial Banking leads banking sector in AI adoption.
AI and Agents
微软 will allow businesses to start making their own autonomous artificial intelligence agents starting next month, in direct competition with @Salesforce, which introduced its own configurable agentic Al tools - Agentforce - in September.
Sequoia Capital released “Generative AI’s Act o1” - a report on how a number of advances over the last year have set the stage to scale “higher-order reasoning” (note the nod to OpenAI’s reasoning model in the report title), claiming ”the agentic reasoning era begins”...
AI Health and Education
英伟达 announced that it will partner with Aidoc , a startup focused on AI tools for healthcare organisations, to develop a novel framework to promote and define AI adoption in the healthcare industry.
Around two-thirds of teachers say they regularly using tools to detect AI-generated content. But, they are not flawless and at that scale, even tiny error rates can add up quickly - with big consequences, says @Bloomberg
@Ellucian, a higher education technology solutions provider, released its second annual AI survey report. This report shows that AI adoption by higher education professionals in the US and Canada for work and personal use has more than doubled over the past year.
Thank you for reading. As you know, feedback always welcome.
And, if you're interested in tailored training in generative AI, please contact me via my friends at Emarketeers: