On Monday, Adobe announced plans to add AI video synthesis tools to Premiere Pro and After Effects, reports Reuters. This follows an announcement last week that AI-powered text-based video editing will come to Premiere Pro in May.

Adobe’s generative video effects will come courtesy of Adobe Firefly. Firefly, an image synthesis mode we previously covered in March, will expand to include video and audio generation tools. Premiere Pro is Adobe’s flagship video editing application, a popular industry tool for assembling TV shows and Hollywood films.

Currently, performing certain special effects or editing tasks can often be labor-intensive. Adobe aims to streamline that process with generative AI assistance. In a demo reel posted on its blog, Adobe showed “Firefly for Video” generating a background soundtrack, generating sound effects, changing the visual season in a video, creating a visual effect on the word “molten,” locating and brightening a person’s face, and generating audio transcripts via text commands.

In its promotional blog, Adobe mentions features we might expect from its new AI tools, including some listed above (like generating sound effects and music), but also text-based color enhancements, text and logo animations, automatic storyboard creation, and creative “co-pilots” that can supposedly generate how-to’s to help users master editing skills.

If these kinds of text-based video editing tools seem familiar, it’s because Runway teased similar capabilities in September, although Runway has yet to deliver on those promises with the fidelity previewed in its teaser. Similarly, it’s probably wise to take Adobe’s marketing video with a grain of salt until the product is released.

(And only time will tell if Adobe Firefly will ever be able to generate a video of Will Smith eating spaghetti that can rival the masterpiece devised by ModelScope, an open source video synthesis model.)

In a separate but related announcement last week, Adobe announced it is integrating text-based video editing into Adobe Premiere Pro, which will allow editors to trim and re-arrange video based on automatically detected transcripts of speech pulled from video clips. Here’s how Adobe describes the feature:

Premiere Pro is the only professional editing software to incorporate Text-Based Editing, revolutionizing the way creators approach their craft by making video editing as simple as copying and pasting text. Powered by Adobe Sensei, Text-Based-Editing automatically analyzes and transcribes clips so editors, producers and assistants can simply copy and paste sentences into any order they want and instantly see them appear on the timeline. Transcripts can be searched in the transcript window to identify exact words and phrases.

Adobe initially previewed a similar text-based video editing capability as part of “Project Blink” in October, although now Adobe says these features come from its “Adobe Sensei” framework. Whether the difference is technical or merely branding is unclear. It’s worth noting that a company called Descript offers a similar feature for video and podcast editing.

Adobe says its Firefly for Video features will begin rolling out “later this year” while also allowing people to sign up to try Firefly, which is currently in closed beta testing. Meanwhile, Adobe promises that text-based video editing is coming to Adobe Premiere Pro in May.

https://arstechnica.com/?p=1932354

L’Intelligenza Artificiale con le sue applicazioni è fra i temi più discussi nel panorama politico e mediatico contemporaneo: come dimostra la recente vicenda legata a ChatGPT e utenti italiani, non passa giorno senza che gli innumerevoli utilizzi dell’AI nei vari campi della vita umana siano protagonisti della cronaca locale o internazionale.

Oggi l’Agenzia dell’Unione europea per la cybersecurity (ENISA), creata quasi 20 anni fa al fine di istituire e rafforzare criteri comuni di sicurezza informatica nell’area comunitaria, affronta organicamente la questione con il report “Cybersecurity of AI and standardisation”.

Pur ammettendo la rilevanza di alcuni elementi hardware o infrastrutturali – nonché il fatto che molti sistemi d’intelligenza artificiale siano ancora in fase sperimentale – l’Agenzia parte dall’assunto “AI is software”, evidenziando come numerose misure e protocolli di software security possano essere applicati anche agli ecosistemi ML/AI. Il modello di riferimento resta infatti la celebre triade “confidentiality, integrity and availability” (CIA), manifestando tuttavia l’intenzione di aggiornarlo alla luce dei nuovi sviluppi tecnici e dell’inedita complessità dei sistemi in esame.

A tale scopo il documento offre una sintesi delle fonti normative e degli standard tecnologici più rilevanti per la materia, alcuni già vigenti (come i noti ISO/IEC 27001 – 27002 – 9001) e altri in divenire, richiamando l’opportunità di proseguire nel solco di una regolamentazione che mantenga la sicurezza come proprio perno centrale.

Prevedere una standardizzazione solida e condivisa in tema di cybersecurity dell’AI è del resto necessario per evitare che le iniziative legislative dedicate, come gli attesi AI Act e Cyber Resilience Act (CRA) in corso d’esame presso le istituzioni comunitarie, finiscano per essere vanificate da lacune o disparità nei sistemi giuridici dei vari Stati UE.

Il report ENISA si pone quindi come un contributo agli sforzi europei verso un quadro regolatorio comune, raccomandando di coinvolgere ogni soggetto interessato – governi, piccole o grandi industrie del comparto tecnologico, standards-developing organisations (SDOs) ed enti di ricerca – nella definizione dei criteri necessari a sviluppare sistemi di AI robusti, affidabili e sicuri.

https://www.ictsecuritymagazine.com/notizie/ai-e-cybersecurity-un-binomio-necessario-secondo-il-nuovo-report-enisa/

Since the launch of OpenAI’s GPT-4 API last month to beta testers, a loose group of developers have been experimenting with making agent-like (“agentic”) implementations of the AI model that attempt to carry out multistep tasks with as little human intervention as possible. These homebrew scripts can loop, iterate, and spin off new instances of an AI model as needed.

Two experimental open source projects, in particular, have captured much attention on social media, especially among those who hype AI projects relentlessly: Auto-GPT, created by Toran Bruce Richards, and BabyAGI, created by Yohei Nakajima.

What do they do? Well, right now, not very much. They need a lot of human input and hand-holding along the way, so they’re not yet as autonomous as promised. But they represent early steps toward more complex chaining AI models that could potentially be more capable than a single AI model working alone.

“Autonomously achieve whatever goal you set”

Richards bills his script as “an experimental open source application showcasing the capabilities of the GPT-4 language model.” The script “chains together LLM ‘thoughts’ to autonomously achieve whatever goal you set.”

Basically, Auto-GPT takes output from GPT-4 and feeds it back into itself with an improvised external memory so that it can further iterate on a task, correct mistakes, or suggest improvements. Ideally, such a script could serve as an AI assistant that could perform any digital task by itself.

To test these claims, we ran Auto-GPT (a Python script) locally on a Windows machine. When you start it, it asks for a name for your AI agent, a description of its role, and a list of five goals it attempts to fulfill. While setting it up, you need to provide an OpenAI API key and a Google search API key. When running, Auto-GPT asks for permission to perform every step it generates by default, although it also includes a fully automatic mode if you’re feeling adventurous.

If tasked to do something like “Purchase a vintage pair of Air Jordans,” Auto-GPT will develop a multistep plan and attempt to execute it. For example, it might search for shoe sellers, then look for a specific pair that meets your criteria. But that’s when it stops because it can’t actually buy anything—at the moment. If hooked into an appropriate purchasing API, that could be possible.

If you want to get a taste of what Auto-GPT does yourself, someone created a web-based version called AgentGPT that functions in a similar way.

Richards has been very open about his goal with Auto-GPT: to develop a form of AGI (artificial general intelligence). In AI, “general intelligence” typically refers to the still-hypothetical ability of an AI system to perform a wide range of tasks and solve problems that are not specifically programmed or trained for.

Like a reasonably intelligent human, a system with general intelligence should be able to adapt to new situations and learn from experience, rather than just following a set of pre-defined rules or patterns. This is in contrast to systems with narrow or specialized intelligence (sometimes called “narrow AI”), which are designed to perform specific tasks or operate within a limited range of contexts.

Meanwhile, BabyAGI (which gets its name from an aspirational goal of working toward artificial general intelligence) works in a similar way to Auto-GPT but with a different task-oriented flavor. You can try a version of it on the web at a site not-so-modestly titled “God Mode.”

Nakajima, the creator of BabyAGI, tells us that he was inspired to create his script after witnessing the “HustleGPT” movement in March, which sought to use GPT-4 to build businesses automatically as a type of AI cofounder, so to speak. “It made me curious if I could build a fully AI founder,” Nakajima says.

Why Auto-GPT and BabyAGI fall short of AGI is due to the limitations of GPT-4 itself. While impressive as a transformer and analyzer of text, GPT-4 still feels restricted to a narrow range of interpretive intelligence, despite some claims that Microsoft has seen “sparks” of AGI-like behaviors in the model. In fact, the limited usefulness of tools like Auto-GPT at the moment may serve as the most potent evidence yet of the current limitations of large language models. Still, that does not mean those limitations will not eventually be overcome.

Also, the issue of confabulations—when LLMs just make things up—may prove a significant limitation to the usefulness of these agent-like assistants. For example, in a Twitter thread, someone used Auto-GPT to generate a report about companies that produce waterproof shoes by searching the web and looking at reviews of each company’s products. At any step along the way, GPT-4 could have potentially “hallucinated” reviews, products, or even entire companies that factored into its analysis.

When asked for useful application of BabyAGI, Nakajima couldn’t come up with substantive examples aside from “Do Anything Machine,” a project build by Garrett Scott that aspires to create a self-executing to-do list, which is currently in development. To be fair, the BabyAGI project is only about a week old. “It’s more of an introduction to a framework/approach, and what’s most exciting are what people are building on top of this idea,” he says.

https://arstechnica.com/?p=1929067

On Wednesday, Databricks released Dolly 2.0, reportedly the first open source, instruction-following large language model (LLM) for commercial use that has been fine-tuned on a human-generated data set. It could serve as a compelling starting point for homebrew ChatGPT competitors.

Databricks is an American enterprise software company founded in 2013 by the creators of Apache Spark. They provide a web-based platform for working with Spark for big data and machine learning. By releasing Dolly, Databricks hopes to allow organizations to create and customize LLMs “without paying for API access or sharing data with third parties,” according to the Dolly launch blog post.

Dolly 2.0, its new 12 billion-parameter model, is based on EleutherAI’s pythia model family and exclusively fine-tuned on training data (called “databricks-dolly-15k”) crowdsourced from Databricks employees. That calibration gives it abilities more in line with OpenAI’s ChatGPT, which is better at answering questions and engaging in dialogue as a chatbot than a raw LLM that has not been fine-tuned.

Dolly 1.0, released in March, faced limitations regarding commercial use due to the training data, which contained output from ChatGPT (thanks to Alpaca) and was subject to OpenAI’s terms of service. To address this issue, the team at Databricks sought to create a new data set that would allow commercial use.

To do so, Databricks crowdsourced 13,000 demonstrations of instruction-following behavior from more than 5,000 of its employees between March and April 2023. To incentivize participation, they set up a contest and outlined seven specific tasks for data generation, including open Q&A, closed Q&A, extracting and summarizing information from Wikipedia, brainstorming, classification, and creative writing.

The resulting data set, along with Dolly’s model weights and training code, have been released fully open source under a Creative Commons license, enabling anyone to use, modify, or extend the data set for any purpose, including commercial applications.

By contrast, OpenAI’s ChatGPT is a proprietary model that requires users to pay for API access and adhere to specific terms of service, potentially limiting the flexibility and customization options for businesses and organizations. Meta’s LLaMA, a partially open source model (with restricted weights) that recently spawned a wave of derivatives after its weights leaked on BitTorrent, does not allow commercial use.

On Mastodon, AI researcher Simon Willison called Dolly 2.0 “a really big deal.” Willison often experiments with open source language models, including Dolly. “One of the most exciting things about Dolly 2.0 is the fine-tuning instruction set, which was hand-built by 5,000 Databricks employees and released under a CC license,” Willison wrote in a Mastodon toot.

If the enthusiastic reaction to Meta’s only partially open LLaMA model is any indication, Dolly 2.0 could potentially spark a new wave of open source language models that aren’t hampered by proprietary limitations or restrictions on commercial use. While the word is still out about Dolly’s actual performance ability, further refinements might allow running reasonably powerful LLMs on local consumer-class machines.

“Even if Dolly 2 isn’t good, I expect we’ll see a bunch of new projects using that training data soon,” Willison told Ars. “And some of those might produce something really useful.”

Currently, the Dolly weights are available at Hugging Face, and the databricks-dolly-15k data set can be found on GitHub.

https://arstechnica.com/?p=1931693

Amazon ha annunciato giovedì il lancio di un proprio servizio di IA generativa attraverso la sua piattaforma di cloud computing, rivolto ai clienti aziendali che desiderano incorporare l’IA nelle loro attività, in modo da separarsi dagli approcci all’IA di altri giganti tecnologici incentrati sui consumatori.

Il servizio di intelligenza artificiale Amazon Bedrock offrirà alle aziende strumenti per costruire e far crescere le proprie applicazioni di intelligenza artificiale progettate per le loro esigenze aziendali, secondo quanto annunciato da Amazon Web Services.

Gli strumenti offerti da Bedrock comprendono due modelli linguistici di grandi dimensioni progettati da Amazon: Titan Text, che è un chatbot simile a ChatGPT in grado di creare testo per cose come post di blog ed e-mail, e Titan Embeddings, che aiuta nella personalizzazione delle ricerche. Quindi l’offerta è più ampia rispetto a ChatGPT. Bedrock includerà anche i modelli delle startup AI21, Anthropic, sostenuta da Google, e Stability AI, che trasforma il testo in immagini, ha dichiarato Amazon.

I clienti potranno personalizzare i modelli Titan in base ai propri dati e alle proprie esigenze, ma le informazioni immesse non “addestrano” i modelli Titan, cioè non entrano a far parte del database di Titan, per cui tutti i dati rimangono protetti da altri clienti e concorrenti, ha riferito CNBC.

Swami Sivasubramanian, vicepresidente di AWS, ha dichiarato alla CNBC che tra le aziende che intendono utilizzare Bedrock ci sono la società di software Pegasystems e la società di tecnologia aziendale Salesforce, proprietaria dell’app di messaggistica aziendale Slack. Quindi presto queste applicazioni potrebbero essere integrate dalla AI.

Amazon ha dichiarato che il suo obiettivo è “democratizzare l’accesso” allo sviluppo dell’intelligenza artificiale, offrendo un servizio online sicuro per consentire alle aziende di creare le proprie applicazioni, che secondo l’azienda sono in grado di eseguire software di intelligenza artificiale in modo più efficiente ed economico rispetto ai concorrenti. Vedremo quali saranno gli effetti sulla comunicazione e sui sistemi aziendali. Già ora molta comunicazione è talmente standardizzata da apparire scritta da una AI.

https://scenarieconomici.it/amazon-presenta-la-propria-ai-bedrock-riservata-alle-aziende/

On Thursday, Amazon released a new suite of AI technologies, including foundational large language models (LLMs) called Titan and a cloud computing service called Bedrock, reports Reuters. The move comes as competitors Microsoft and Google integrate AI chatbots into their search engines and cloud operations.

Recently, LLMs like OpenAI’s GPT-4 and its instruction-tuned cousin ChatGPT have become one of the hottest stories in tech, inspiring large investments in AI labs and shaking up business operations in major players. LLMs, often grouped with similar technologies under the umbrella term “generative AI,” can take any kind of written input and transform, translate, or interpret it in different ways.

In response, Amazon’s cloud computing division, Amazon Web Services (AWS), has designed its new AI technologies to help companies develop their own chatbots and image-generation services (such as OpenAI’s DALL-E). As with Microsoft’s Azure cloud platform that powers OpenAI’s models, Amazon stands to reap financial rewards from renting out the computing muscle to make its own brand of generative AI happen.

AWS’s core offering, named Bedrock, enables businesses to customize “foundation models” using their own data. In AI, foundation models are core AI technologies that serve as starting points for companies to build upon. Using fine-tuning (additional training with a specific goal), companies can incorporate proprietary data, customizing the models for their own needs. OpenAI offers a similar service that allows customers to fine-tune models for custom chatbots.

Bedrock will provide customers with Amazon’s proprietary LLM foundation models, known collectively as Amazon Titan, as well as models from third-party companies. Startups AI21 Labs, Anthropic, and Stability AI will offer their models alongside Amazon’s.

According to Amazon, one key aspect of providing its Bedrock service will be allowing AWS customers to test these new AI technologies without managing the underlying data center servers that power them. Those data centers are costly investments. Amazon says the underlying servers for the Bedrock service will utilize a combination of Amazon’s custom AI chips (AWS Trainium and AWS Inferentia) and GPUs from Nvidia, which is currently the largest supplier of AI chips.

Also on Thursday, Amazon announced a preview of Amazon CodeWhisperer, an AI-powered coding assistant similar to GitHub Copilot and Replit Ghostwriter. It’s free for individual use and available for evaluation today.

https://arstechnica.com/?p=1931614

President Joe Biden’s administration wants stronger measures to test the safety of artificial intelligence tools such as ChatGPT before they are publicly released, though it hasn’t decided if the government will have a role in doing the vetting.

The U.S. Commerce Department on Tuesday said it will spend the next 60 days fielding opinions on the possibility of AI audits, risk assessments and other measures that could ease consumer concerns about these new systems.

“There is a heightened level of concern now, given the pace of innovation, that it needs to happen responsibly,” said Assistant Commerce Secretary Alan Davidson, administrator of the National Telecommunications and Information Administration.

The NTIA, more of an adviser than a regulator, is seeking feedback about what policies could make commercial AI tools more accountable.

Biden last week said during a meeting with his council of science and technology advisers that tech companies must ensure their products are safe before releasing them to the public.

{ Feature: ChatGPT, the AI Revolution, and the Security, Privacy and Ethical Implications }

The Biden administration also last year unveiled a set of far-reaching goals aimed at averting harms caused by the rise of AI systems, but that was before the release of ChatGPT, from San Francisco startup OpenAI, and similar products from Microsoft and Google led to wider awareness of the capabilities of the latest AI tools that can generate human-like passages of text, as well as new images and video.

“These new language models, for example, are really powerful and they do have the potential to generate real harm,” Davidson said in an interview. “We think that these accountability mechanisms could truly help by providing greater trust in the innovation that’s happening.”

The NTIA’s notice leans heavily on requesting comment about “self-regulatory” measures that the companies that build the technology would be likely to lead. That’s a contrast to the European Union, where lawmakers this month are negotiating the passage of new laws that could set strict limits on AI tools depending on how high a risk they pose.

Related: Cyber Insights 2023 | Artificial Intelligence

Related: White House Unveils Artificial Intelligence ‘Bill of Rights’

Related: Bias in Artificial Intelligence: Can AI be Trusted?

Related: The Starter Pistol Has Been Fired for Artificial Intelligence Regulation in Europe

https://www.securityweek.com/biden-administration-seeks-input-on-ai-safety-measures/

Two months back, we launched Buffer’s AI Assistant.

It’s one of the most popular features we’ve ever launched at Buffer.

Since launching AI has assisted in creating over 31,500 posts for 8,100 users.

That’s 2.5 million AI-generated words created in Buffer.

So, you won’t be surprised to hear we’ve been hard at work making it both smarter and faster.

From today, Buffer’s AI Assistant can:

• Rephrase your posts
• Summmarize long-form text
• Expand sentences into paragraphs
• Generate personalized social media ideas

Plus, it’s integrated with Canva’s AI image generator—helping you create the perfect picture for your post.

Assistance right when you need it

If you’ve ever struggled to come up with fresh ideas for your social media content, Buffer’s AI Assistant can help. By answering questions about your brand and target audience, you can generate new ideas that are perfectly suited for your account.

Ideas are one thing, but ultimately you need to tweak those ideas until they’re ready to publish.

To help the AI Assistant also comes equipped with features that help you rephrase, summarize, and expand your posts.

This means that you can take existing content and make it more engaging and effective without spending hours writing and rewriting.

Lightning-fast content creation

So, how does this help you create better content for social media, faster?
For starters, the idea generation feature saves you time and brainpower. Instead of wracking your brain for hours trying to come up with new ideas, you can let the AI do the heavy lifting.

This frees up your time to focus on other aspects of your social media strategy, like crafting compelling visuals or engaging with your followers.

The rephrasing, summarizing, and expanding features are also incredibly helpful. If you’re struggling to find the right words to convey your message, the AI Assistant can help you find a new way to phrase things.

If you need to summarize a longer piece of content, the AI can help you distill the main points down to their essence. And if you need to expand on a particular topic, the AI can help you find new angles and fresh insights.

How it stacks up

When we launched Buffer’s AI Assistant back in February 2023, we were one of the first to market. Today, many tools like Buffer have some form of AI available.

So what makes Buffer stand out?

First up, our AI Assistant is available across every channel. This means you can use the AI Assistant to improve your content on every social media platform you use.

Secondly, the AI-powered personalized social media ideas aren’t available elsewhere. Simply answer a few simple questions about your brand and target audience, and voila, you’ve got endless ideas to try.

Plus, the AI Assistant is available directly in the Buffer composer. This means you can create, edit, and improve your content all in one place, without having to switch between different tools or windows.

Let’s face it, the best way to learn is to test it out yourself.

So, don’t wait around, try Buffer’s AI Assistant for free →

https://buffer.com/resources/why-were-doubling-down-on-ai-at-buffer/

China proposed new checks on artificial intelligence chatbots on Tuesday, in an effort to control how its tech industry rolls out generative AI models, as popularized by ChatGPT in the US.

Hours after tech giant Alibaba followed its peers SenseTime and Baidu with the launch of a ChatGPT-like bot, China’s powerful Internet regulator released draft measures likely to slow Alibaba’s rollout, citing chatbots’ potential for “social mobilization.”

The Cyberspace Administration of China proposals said providers would have to submit their products for security reviews before their public release and it would set up a database to register them. The regulator also said platforms must verify users’ identities, allowing usage to be tracked.

“Content generated by generative artificial intelligence should embody core socialist values and must not contain any content that subverts state power, advocates the overthrow of the socialist system, incites splitting the country or undermines national unity,” the CAC rules state.

Earlier on Tuesday, Alibaba Chief Executive Daniel Zhang said Alibaba’s Tongyi Qianwen, which roughly translates to “truth from a thousand questions,” would be incorporated as an AI bot in workplace collaboration tool DingTalk and the company’s Tmall Genie smart speakers before being integrated into all its products at some point in the future.

“We are at a technological watershed moment driven by generative AI and cloud computing,” Zhang said, as Alibaba became the latest Chinese company to declare it would try to emulate US start-up and ChatGPT creator OpenAI.

“Ten to 20 years from now, when we look back, we will realize we were all on the same starting line,” Zhang said. “Seizing these future opportunities is our common wish and requires a shared vision.”

Xu Li, chief executive of Chinese AI software leader SenseTime, unveiled its SenseChat bot on Monday with a live demonstration, when it was able to write an email and tell a story about a cat catching a fish.

Baidu last month unveiled its chatbot, Ernie, to mixed reviews. Users said the chatbot could create vivid pictures, but it struggled with basic logic and was unable to write code like OpenAI’s GPT-4 could.

Alibaba’s efforts are tucked inside its cloud arm, which Zhang personally took control of in December amid slumping growth. The chief is betting that a radical restructuring of the group announced last month will help reset the company’s fortunes. Alibaba will split into six entities but will remain controlled by a common holding company, allowing each unit to be more nimble.

Tongyi was opened to corporate clients for testing last Friday. Early user feedback posted online showed the chatbot could write poems in Chinese and French and solve basic mathematical problems but struggled with simple logic.

When one user asked Tongyi how to stir-fry reinforced concrete to make a tasty dish, it offered a recipe that included slicing the concrete into small pieces. Baidu’s Ernie similarly told users to mix concrete with garlic, onions, and peppers, and noted concrete was “a very special ingredient with a unique texture.”

When ChatGPT was asked the same question it said it did not understand the request and that concrete was not edible. Other Chinese user tests showed Ernie and Tongyi making up facts about non-existent people. CAC’s new rules would prohibit the chatbots from generating false information.

The Financial Times was able to replicate the queries and problematic answers in Baidu’s Ernie but did not have access to Tongyi.

Ernie requires users to provide their names and national ID numbers for access. The chatbot returns a brief biography of Xi Jinping for any question related to the Chinese president.

The flurry of generative AI announcements has helped boost Chinese tech groups’ stock prices. Baidu’s shares have risen more than 15 percent from the start of the year, while SenseTime is up nearly 50 percent. Alibaba’s shares rose 1.5 percent in Hong Kong on Tuesday.

The gains in AI stocks have alarmed Chinese regulators and state media, which have repeatedly warned investors against chasing the speculative frenzy.

© 2023 The Financial Times Ltd. All rights reserved. Not to be redistributed, copied, or modified in any way.

https://arstechnica.com/?p=1930659

Since last year, a group of artists have been using an AI image generator called Midjourney to create still photos of films that don’t exist. They call the trend “AI cinema.” We spoke to one of its practitioners, Julie Wieland, and asked her about her technique, which she calls “synthography,” for synthetic photography.

The origins of “AI cinema” as a still image art form

Last year, image synthesis models like DALL-E 2, Stable Diffusion, and Midjourney began allowing anyone with a text description (called a “prompt”) to generate a still image in many different styles. The technique has been controversial among some artists, but other artists have embraced the new tools and run with them.

While anyone with a prompt can make an AI-generated image, it soon became clear that some people possessed a special talent for finessing these new AI tools to produce better content. As with painting or photography, the human creative spark is still necessary to produce notable results consistently.

Not long after the wonder of generating solo images emerged, some artists began creating multiple AI-generated images with the same theme—and they did it using a wide, film-like aspect ratio. They strung them together to tell a story and posted them on Twitter with the hashtag #aicinema. Due to technological limitations, the images didn’t move (yet), but the group of pictures gave the aesthetic impression that they all came from the same film.

The fun part is that these films don’t exist.

The first tweet we could find that included the #aicinema tag and the familiar four film-style images with a related theme came from Jon Finger on September 28, 2022. Wieland, a graphic designer by day who has been practicing AI cinema for several months now, acknowledges Finger’s pioneering role in the art form, along with another artist. “I probably saw it first from John Meta and Jon Finger,” she says.

It’s worth noting that the AI cinema movement in its current still-image form may be short-lived once text2video models such as Runway’s Gen-2 become more capable and widespread. But for now, we’ll attempt to capture the zeitgeist from this brief moment in AI time.

Julie Wieland’s AI story

To get more of an inside look at the #aicinema movement, we spoke to Wieland, who’s based in Germany and has racked up a sizable following on Twitter by posting eye-catching works of art generated by Midjourney. We’ve previously featured her work in an article about Midjourney v5, a recent upgrade to the model that added more realism.

AI art has been a fruitful field for Wieland, who feels that Midjourney not only gives her a creative outlet but speeds up her professional workflow. This interview was conducted via Twitter direct messages, and her answers have been edited for clarity and length.

Ars: What inspired you to create AI-generated film stills?

Wieland: It started out with dabbling in DALL-E when I finally got my access from being on the waitlist for a few weeks. To be honest, I don’t like the “painted astronaut dog in space” aesthetic too much that was very popular in the summer of 2022, so I wanted to test what else is out there in the AI universe. I thought that photography and movie stills would be really hard to nail, but I found ways to get good results, and I used them pretty quickly in my day-to-day job as a graphic designer for mood boards and pitches.

With Midjourney, I reduced my time from looking for inspiration from Pinterest and stock sites from two days of work to maybe 2–4 hours, because I can generate the exact feeling I need, to get it across for clients to know how it will “feel.” Onboarding illustrators, photographers, and videographers has never been easier ever since.

https://arstechnica.com/?p=1904461