• Can AI commit electric plagiarism?
• Can use of generative AI to approximate a writer’s style be considered as plagiarism?
• Would authors prefer to be pirated than plagiarized?

AI plagiarism might be about to take Z library’s place as the plague of the literary world.

Earlier this year, when Z library was still making headlines, authors were united in the battle against piracy. Particularly vocal was the Authors Guild, which recently came to the defence of Professor Jane Friedman, an author who specializes in helping other writers get published.

She took to the-platform-formerly-known-as-Twitter to decry Amazon after she discovered that books she didn’t write were being attributed to her.

Friedman’s tweet and article detailing the issue.

Over the last 25 years, Friedman has written or contributed to 10 books on the industry. However, she hasn’t published anything new since 2018. When a reader reached out to her about more recent works, alarm bells rang.

Titles of what Friedman called “garbage books” included Your Guide to Writing a Bestseller eBook on Amazon, Publishing Power: Navigating Amazon’s Kindle Direct Publishing, and Promote to Prosper: Strategies to Skyrocket Your eBook Sales on Amazon.

Friedman’s solid following is based on her work which includes similar titles like The Business of Being a Writer, What Editors Do, and Publishing 101.

When she contacted Amazon to get the faked books removed, the e-commerce giant refused, even though they were being traded on the basis of her name and reputation.

READ NEXT

Forcing shadow libraries out of the darkness

Because Friedman didn’t hold a trademark to her own name, she couldn’t get a straightforward copyright infringement case.  When she filed a report with Amazon, its response was to ask for “any trademark registration numbers that relate to your claim.”

In an article from Plagiarism Today, which Friedman retweeted, it’s pointed out that historically, authors have had two key battles: piracy and plagiarism. Both of these fall under copyright law – which is why Z library has been involved in legal proceedings.

However, a third issue has arisen out of the advent on generative AI. What Friedman is fighting can be described as “reverse plagiarism.”

Plagiarism? Or identity theft for profit?

Amazon’s stance was that unless the books copy text or other protectable elements from Friedman’s work, copyright doesn’t apply to these cases. Even though the books are likely AI-generated works based on Friedman’s content, there’s nothing to sustain a copyright infringement claim.

“We have clear content guidelines governing which books can be listed for sale and promptly investigate any book when a concern is raised,” Amazon spokesperson Ashley Vanicek told Decrypt by email. “We welcome author feedback and work directly with authors to address any issues they raise, and where we have made an error, we correct it.”

Online, other authors shared similar stories. The Authors Guild also showed support.

The Authors Guild response to Friedman’s original Tweet.

Also speaking to Decrypt, the organization said, “We’ve worked with Amazon on this issue in the past, and we will continue our conversations with it about advancing its efforts to keep up with the technology.”

“Meanwhile, we encourage everyone to report these books that try to profit from your brand through Amazon’s complaint portal.”

The global infatuation with AI has been a sticking point for writers this year already. Film and television productions ground to a halt when the members of the WGA went on strike after negotiations with the Alliance of Motion Picture and Television Producers collapsed in May. When 160,000 members of SAG-AFTRA also went on strike, their concerns were similar.

The capabilities of AI have caused many to wonder whether there’ll be jobs available to humans once its full capabilities are harnessed. For writers like Friedman, a future in which AI replaces people is fast becoming a reality.

YOU MIGHT LIKE

ARTIFICIAL INTELLIGENCE

Can artificial intelligence have ethics?

Friedman wrote that she thinks “her” books were AI generated because “I’ve used these AI tools extensively to test how well they can reproduce my knowledge. I also do a lot of vanity prompting, like “What would Jane Friedman say about building author platform?”

In July, 10,000 members of the Authors Guild co-signed a letter calling on AI industry leaders to obtain consent from, credit, and fairly compensate authors.

The Guild also submitted written testimony to the Senate Intellectual Property Subcommittee for its July 12 hearing on artificial intelligence, underscoring the threat to “the written profession from unregulated use of generative AI technologies that can produce stories, books, and other text-based works and displace the works of human authors in the marketplace.”

Ironically, given the firm stance that many authors took against Z library’s provision of free books, Friedman’s take is that she “would rather see [her] books get pirated than this.”

Jane Friedman. Source: janefriedman.com

The attention that Friedman’s tweet got meant that Amazon removed the books from its site. They’re now listed as not available to buy.

The fake books are still on Goodreads under Friedman’s name though, increasing her concern that AI generated work is going to ruin the credibility of authors’ real work.

The question of whether use of generative AI to mimic an author’s style counts as plagiarism has been debated by colleges and schools that worried students would use tools like ChatGPT to write assignments.

Further, because the Large Language Models that generative AI bots run on are trained on existing work, there’s an argument to be made that any content created by AI is, on some level, plagiarized. That’s an argument familiar from the world of art, where generative AI programs commonly take elements of human work and repurpose them without credit, premission, or remuneration.

Maya Shanbhag Lang, president of the Authors Guild, said, “The output of AI will always be derivative in nature. AI regurgitates what it takes in, which is the work of human writers. It’s only fair that authors be compensated for having ‘fed’ AI and continuing to inform its evolution. Our work cannot be used without consent, credit, and compensation. All three are a must.”

READ NEXT

What are the risks of artificial intelligence without ethics?

For now, Friedman is focusing on what she can control: her own writing. She said she’s revisiting her book the Business of Being a Writer.

 “At least now I will have a good story to include.”

Plagiarism – it’s not new, but it is getting cleverer.

The post AI plagiarism: Z library no longer the biggest battle authors face appeared first on TechHQ.

• Nestlé is using AI to provide optimized marketing campaigns.
• But the brand remains tainted by the ‘baby milk scandal.’
• AI tools can work to both improve and destroy a brand’s reputation.

Don’t panic, your chocolate isn’t going to be turned into a virtual experience just yet, but Nestlé has officially joined the ranks of companies that are integrating AI into their businesses in one way or another.

In February, the food and drinks titan announced that its 15,000-person-strong marketing division would utilize AI tech to help run campaigns across social platforms while maintaining brand consistency. This meant that creative assets produced for any of its 200 brands must be run through an AI system trained on its strict marketing do’s and don’t’s.

The system, named Cortex, comes from creative consultancy CreativeX, which states on its website that “Brand consistency has been shown time and time again to drive lift across both brand and sales KPIs.” An attractive prospect for Nestlé, which hasn’t had the smoothest ride regarding company reputation. In short, it needs to improve its image, consistently.

YOU MIGHT LIKE

BLOCKCHAIN

Nestlé – why the world’s biggest food company uses blockchain

It took its largest and most notorious blow about 50 years ago when campaign groups started highlighting what’s now known as the ‘Nestlé baby milk scandal’. In the 1960s, many infant formula companies shifted their campaigning to developing countries to depict it as “scientific, modern, prestigious, and (falsely) nutritionally superior to breastmilk.” Nestlé sent salespeople out to countries in Latin America, Asia, and Africa to promote their products dressed as so-called “milk nurses”.

Nestlé Milk nurses in South Africa. Source: Crowther S M et al. (2009)

They pushed the line that this expensive formula would be more beneficial to children than what their mothers’ bodies produced, which is untrue. Nestlé was later accused of providing incentives to doctors to promote their products.

Many mothers in these developing countries diluted the formula to make it last, often with dirty water, which reduced its nutritional value. As a result, many infants suffered from malnutrition, dehydration, and various illnesses, leading to a higher-than-necessary infant mortality rate.

The aggressive marketing tactics that Nestlé and associated formula companies employed are said to have contributed to a decline in breastfeeding rates in these regions, where it was once the norm and a crucial source of nutrition and protection against diseases.

After health organizations began raising concerns about these unethical marketing practices, Nestlé’s reputation took a turn, and it received significant backlash and boycotts from consumers.

Over time, the company has attempted to address the criticism and improve its image by adopting changes in its marketing strategies. Nestlé now adheres to the International Code of Marketing of Breast-milk Substitutes, created by the World Health Organization and UNICEF in 1974, but activists and critics continue to closely monitor its practices to ensure compliance.

Nestlé now claims to adhere to the International Code of Marketing of Breast-milk Substitutes. Source: Shutterstock

The AI tool Nestlé marketers are using does not claim to be a reputational tonic. Aude Gandon, Nestlé’s senior vice-president and global chief marketing and digital officer, hopes instead that it will change the brand’s advertising focus from TV to digital-first.

When CreativeX was first recruited in 2021, its team trained an AI on thousands of previous Nestlé campaigns and their results. This meant it could learn the elements of these ads that drove the best ROI. Once the learning period was over, Cortex was able to analyze any creative asset and give it a ‘Creative Quality Score’ based on how many elements it contains which prove successful on a given social platform, like YouTube or Facebook. These include aspect ratio, video length, subtitles, and calls to action.

Staff must now put any asset through the AI before it goes live to ensure it is optimized for their chosen platform. According to Ms Gandon, this is to stop time-wasting during marketing meetings where attendees may disproportionately fret about whether minute details, like logo placement, impacted the campaign’s failure.

While the AI-determined rules may sound “unsexy”, as CreativeX’s chief executive Anastasia Leng told The Drum, they are not intended to replace the creative input of the marketers. “These are basic things that create the canvas on which an idea can be successful,” she said.

Cortex is not the only way Nestlé has integrated AI into its business. In 2021 it launched a ‘Cookie Coach’ named Ruth, a virtual avatar that could answer basic questions about its Toll House chocolate chip cookie recipe. Ruth was born from the number of questions on how to recreate these cookies that Nestlé received during the COVID-19 pandemic.

In 2021, Nestlé launched a ‘Cookie Coach’ named Ruth, a virtual avatar that could answer basic questions about its Toll House chocolate chip cookie recipe. Source: Nestlé

Research from Greyb also revealed that the company has filed patents for AI technologies that could recommend personalized dietary requirements. Nestle Health Science offers various nutritional supplements, like Boost Women, and provides data for screening and assessment tools that can recommend these products. These tools currently work simply by making recommendations based on the user’s responses to a questionnaire, but the patents suggest these could soon be further optimized with AI.

But could marketing teams use AI to repair a brand’s reputation? These days, many companies make use of AI-driven sentiment analysis to monitor online discussions, social media platforms, and customer reviews for negative opinions. These ensure that representatives can address any concerns in a timely manner and demonstrate that customer satisfaction is a priority.

Similarly, AI-driven moderation tools can be employed to identify and remove harmful or misleading online content and help preserve company reputation. It can also help prevent future scandals by detecting the initial indicators of an escalating situation, allowing brands to take immediate action. Many PR companies now offer AI tools as part of their packages.

Creatively integrating the latest AI-powered innovations could also provide the opportunity for a reputational clean slate. By introducing a never-before-seen feature or service that genuinely addresses customer needs and enhances their experience, a brand can shift the narrative away from past scandals and redefine the company’s image as an AI trailblazer.

Although AI can help with marketing campaigns and safeguarding brand reputation, it cannot fully replace human judgment and creativity. Source: Shutterstock

An example of this is how OpenAI itself, the creator of ChatGPT, was outed as paying workers in Kenya a maximum of $2 an hour to label violent, sexist, and racist text data as such so that it could be used to train an AI-powered safety mechanism for ChatGPT. Not long after, OpenAI released GPT-4, its most advanced system yet, to the public.

As much as advances in AI can help a brand’s reputation, they can also work against it. An example of this is the Cambridge Analytica scandal from 2018, where the data analytics firm misused advanced AI-driven techniques to harvest and exploit the personal data of millions of Facebook users without their consent.

This data mining operation aimed to build detailed psychographic profiles of individuals, which were then used to deliver targeted political advertisements and influence the outcome of elections, including the 2016 US Presidential election and the Brexit referendum.

YOU MIGHT LIKE

SOCIAL MEDIA

Facebook fined maximum possible in UK over Cambridge Analytica affair

While an increasing number of companies embrace AI to enhance the efficiency and effectiveness of their products and services, they also open themselves up to a host of potential controversies related to its use. While these may not be as high profile as what happened with Cambridge Analytica, a wayward chatbot or biased algorithm can inflict significant reputational damage.

The majority of AI tools can’t fact-check information, so companies that rely on generated content put themselves at risk of associating inaccurate information with their brand. Gartner predicts that 80 percent of marketers will deal with content authenticity issues by 2027.

What’s more, AI-derived copy can lack originality and personality, as it is produced by an algorithm that only works by re-assembling oft-assembled words and sentences. That tends to produce results that are safely generic. Chatbots are also open to manipulation and to proliferate bias from their training data, so aligning one with a company comes with a certain degree of danger.

As Ms Leng put it, many creatives are concerned that AI tools can only produce “cookie cutter” content. It is true that, although AI can help with marketing campaigns and safeguarding brand reputation, it cannot fully replace human judgment and creativity. Therefore it is crucial for companies like Nestlé to tread carefully, ensuring a balance between harnessing AI’s potential and preserving the essence of the human touch, if it wants to remain a household name.

The post Sweetening things up: Can AI help repair Nestlé’s reputation? appeared first on TechHQ.

The revelation that Meta’s tracking technology, Pixel, was deployed on three US tax return preparation websites has caused shockwaves that have spread as far as the Senate. A report by Democrats urges further investigation to see exactly what information Meta had access to.

In a letter to the IRS and several other peri governmental organizations, the seven signatories say there has been “a shocking breach of taxpayer privacy by tax prep companies and by Big Tech firms.”

The tax preparation companies installed Pixel code snippets on their websites that allowed them to monitor users’ activities while on site. The data is sent to Meta, which correlates it and helps companies optimize their activities for marketing or site optimization.

It’s worth noting that the tax-preparation companies also ran similar code snippets from Google, which denied tracking users.

Meta Pixel’s tracking

What’s most surprising about the “revelations” is the revelation that many people are surprised. Tracking end-users is commonplace, bordering on ubiquitous on the modern ‘web. Whether using a browser or mobile app, internet users are constantly tracked through Meta’s Pixel, Google Analytics cookies, or any number of the many thousands of tracking methods.

On this author’s smartphone, for example, there have been 29,313 tracking attempts recorded in the last week. A tracking attempt typically comprises third-party software installed in an app (or website) attempting to “phone home” with data such as location, network, phone ID, ZIP code, email address, contacts lists, and many more juicy digital tidbits.

Source: Fosstodon.org

That situation has led to the emergence of many ad-blocking, anti-tracking and -fingerprinting methods, including browser add-ons such as UBlock Origin, Privacy Badger, and NoScript. A game of cat-and-mouse is constantly played out by digital advertisers and anti-tracker software developers, with new methods of fingerprinting users springing up as quickly as prophylactic methods are spun up.

Tracking technology is deliberately simple to deploy on an organization’s internet real estate (websites and apps) and is often free to use. Data is collected by the third party and can be used for its own purposes. In addition to Meta, there’s Google, Adobe, OneSignal, Microsoft, Urban Airship, Criteo, Amazon, Index Exchange, Bing, Improve Digital, Adform, Yahoo, Twitter, Zemanta, Yieldlab, et al. ad nauseam – there are literally thousands of companies offering tracking methods.

Meta Pixel in black and white

The small print of the Pixel documentation does note that some users may need to be wary of GDPR legislation, and those wishing to collate data from iOS devices may struggle due to Apple’s shutting down of default tracking capabilities on apps available from its App Store.

Source: Meta

The horror exhibited in the tone of the Democrats’ lawmakers’ letter to the IRS and its watchdog exhibits the kind of naivete that is all too prevalent. Similar tones of outrage are present when journalists “discover” that TikTok (owned by Bytedance) allows access to American and Australian citizens’ data by Chinese people working for a Chinese company.

Source: Buzzfeed

Source: The Guardian

The truth is that any company deploying tracking technology for whatever reason on its website or in its apps is sending data to the company that supplies the tracking code. If an organization works in any area where privacy is important to its users, it must know that its real estate is handing information to a third party.

Although companies may only be interested in their customers’ traversals around their websites, the data collected by the tracking technology company may not be limited. Similarly, those signing up for the “free” tiers of user tracking may get only limited metrics (until they start to pay up, of course). But the tracking company – you can be sure – will absorb all the information it can.

That a third party receiving data may be in Beijing or San Francisco is irrelevant. Companies need to know that using off-the-shelf tracking technology supplied by a third-party spills their information to that third party. Whether that’s a good deal to get internal marketing insights is highly debatable.

The post Meta Pixel “scandal” surprises too many tax-payers appeared first on TechHQ.

In May 2023, Adobe released a tutorial on how to use text-based video editing in Premier Pro, having added the feature to its flagship software. And the official arrival (previously, there were workarounds using XML imports) of this breakthrough way of editing video in leading content creation software shows how popular text-based video (and audio) editing is becoming.

The amount of time required to turn raw footage into a finished video depends on a number of factors, such as the complexity of the project and the talents of the production crew. But, as a rule of thumb, editors can expect to spend anywhere from 30-60 minutes on a project for every minute of the finished video. Or at least that used to be the case.

What is text-based video editing?

As content creators are discovering, text-based video editing – which allows users to edit a video as if they were editing a document (and it works for audio too, if you are making a podcast) – dramatically speeds up film production and splicing up clips for use on social media.

Videographers (or podcasters) can have a rough cut of their work in minutes rather than hours. And for those that are new to editing videos or spoken audio works, being able to use familiar document editing skills makes it easy to jump up the multimedia learning curve.

And how does text-based video editing work?

AI transcription speech-to-text services have been around for some time (and applied to various industrial use cases), and their success paves the way for text-based video editing. Adobe’s Premier Pro integration gives a good overview of how the process works. Firstly, AI speech-to-text algorithms transcribe the source footage. And when complete, the text appears in a transcript window.

Cleverly, because the text has been matched to the video, users can edit the video timeline by simply moving words or phrases rather than having to cut and paste clips. Typically, users find text-based video editing a much more intuitive process and can focus their efforts on the content rather than having to worry about the technicalities of splicing footage.

READ NEXT

Generative AI can read your mind, if you let it

Even seasoned video editors, writing on user forums, talk about how creating text-based video edits has changed their life. Some even go as far to say that being able to perform text-based video editing natively within a non-linear editing (NLE) system is one of the most significant advancements in digital editing. And the concept has been gathering pace for a while.

In fact, the idea of editing film and audio as if you were editing a text document has been around for over a decade. And tools have existed for a similar period of time – for example, prEdit has long allowed users to make subclips from transcribed media files and send the story to Final Cut Pro or Premiere Pro as a cut sequence ready for further editing.

Very interesting to see everyone all of a sudden release text based video editing even though it was possible for so long already.

Funny how AI stuff was here for a while but we were all caught up in Web 3.0 whatever that was

— Matti Haapoja (@MattiHaapoja) April 17, 2023

What’s changed is the growth and capability of cloud-based speech-to-text transcription services that fully automate spoken audio processing and support a variety of languages. And the availability of a wide number of apps means that you don’t need to be a professional filmmaker, or even own a copy of Adobe Premier, Final Cut Pro, or DaVinci Resolve to benefit from text-based video (and audio) editing.

And there are a ton of ways that having auto-generated, time-matched transcriptions helps content creators. Users can very quickly search for topics rather than having to scroll through and listen to the audio. Text-based video editing makes it easy to remove any so-called disfluencies or hesitations in the dialogue, such as mentions of ‘um’ or ‘er’, vocal mistakes, and any overused filler phrases – like, you know, that kind of thing.

Solutions can be air-gapped too – for example, if you’re working on sensitive interviews, confidential recordings, or a blockbuster movie that you want to keep under wraps until its official launch date.

Synthetic voice: no need for microphone overdubs

Also, app users don’t have to give up on pro features. And there are some interesting software additions too – for example, Descript and Simon Says allow users to collaborate remotely on text-based multimedia editing.

READ NEXT

Text-to-speech: exploring generative AI’s talkative side

Descript’s podcast studio tool moves the needle further by integrating synthetic voice capabilities. Users can delete an unwanted word or phrase in the time-synchronized transcript, type new dialogue, and the text-based editor will synthesize and insert the audio content to match the existing voice track.

On the video image side of things, AI algorithms can even digitally rotate a speaker’s eyeballs (link to video demo) so that the subject can (in real life) be looking down, reading from a script, but (in video) appear to be looking at the camera. And this really only begins to touch on what’s possible.

Opinions vary, but text-based video editing can be anywhere from 5 – 12 times faster than using conventional methods. In principle, as Milk Video’s creative team points out, trimming scenes using a transcript allows users to edit video at the speed of reading. And for organizations that produce a lot of video content, training materials, podcasts, and other multimedia, that’s a big time-saving.

12 best text-based video (and audio) editing apps

1. Descript
2. Streamlabs Podcast Editor
3. Riverside
4. Trint
5. Simon Says
6. Camtasia Audiate
7. Sonix AudioText Editor
8. Kapwing
9. Milk Video
10. Imvidu
11. Pictory
12. Reduct Video

There are some great community projects too, and examples include the text-based video editing proof of concept built by Radamés Ajna and shared on Hugging Face.

When big players such as Adobe make text-based video editing a standard feature, you can be sure that the landscape has changed. And once content makers start using these tools, they won’t want to give those time-savings back.

The post Tech tools for business: best text-based video editing apps appeared first on TechHQ.

• Discord to launch server subscriptions.
• Paywalls will monetize certain parts of the service.
• Discord’s brand of subverting the norm in peril.

Discord is introducing new monetization features by expanding server subscriptions, a Patreon-like subscription button that’s been available on large servers since 2022. The site is introducing tiered subscriptions and longer-term plans to effectively turn servers into storefronts.

“Effectively storefronts” Discord server subscriptions. Via: https://discord.com/blog/server-subscriptions-updates-media-channels-tier-templates-and-more

In December 2022, server subscriptions launched as a way for community leaders to make money directly on Discord “by providing exclusive experiences, available only for their paying subscribers.”

In a blog post titled “Server subscriptions just got super powered,” Derek Yang wrote that “To date, we’ve paid out millions of dollars to thousands of creators and communities, and we’re seeing more creators and communities earning on their Discord servers every day… Today we’re excited to share new tools that help you get started earning money faster.”

READ NEXT

Meta readying to monetize digital assets in the metaverse

Discord server subscriptions – super powered?

This week, “media channels,” designed to host subscriber-only content, was launched in beta. The suggestion is that the feature could be used for “exclusive memes and wallpapers,” or that artists could monetize their work by putting it behind a paywall. This might sound familiar – it’s how Patreon functions already.

Yang says paywalled media channels are a way to “give your subscribers lavish insider content” in addition to what the server already offers for free.

Other monetization opportunitites on offer are:

• Tier Templates: Formalized subscription tiers with prices set by Discord ($3.99, $4.99, $7.99, and $9.99)
• Downloadables: One-time purchasable digital products or subscriptions sold by server owners, which will be accessed via…
• Server Shops: “A single home for server owners to sell Server Subscriptions, Downloadables and Premium Roles.”

Via https://discord.com/blog/server-subscriptions-updates-media-channels-tier-templates-and-more

In the announcement, there’s a reminder that “not every opportunity to generate revenue needs to become a get-rich-quick” scheme. The emphasis is very much on “a little extra money to put back into making your community even better.”

Which is ironic, given Discord plans to take a 10% cut of server subscriptions which – call us pessimistic and cynical – is unlikely to go back into the community. By encouraging users to lock conversations behind paywalls, Discord is sharing tips to “convert people into paying customers.”

For PC Gamer, Morgan Park writes that the changes make him “miss the days when it felt like Discord was making Discord for [users].”

READ NEXT

Reddit protests cause blackouts onsite

An exercise in pandering?

Discord is at heart a chat platform for groups to discuss common interests. It provides an easy way to talk with friends online, and servers are often used in a similar way to a group chat. However, Discord is increasingly pandering to large communities with the introduction of features like live stages and forum channels.

The issue is that the transition from user to consumer is tricky, especially in terms of what it means for a moderator. They’ll be given free rein to paywall individual functions of a server, as Discord doesn’t plan to police what server owners charge.

Yang draws attention to Valorant streamer Woohoojin’s Club Banana server as a successful Discord community funded by subscriptions.

Woohoojin, who makes over $16,000 every month from Discord subscriptions, hands out free subscriptions to regular viewers and only asks viewers to give him money if “they have lots of it.”

The story isn’t new: over the last decade, tech companies have begun revoking the subsidized products that garnered enough attention to gain a monopoly before slamming user fees up and cashing out (see Netflix clampdowns on password sharing or the whole thing with Twitter Blue).

What’s different here is that Discord has built a userbase by subverting the norm. It didn’t inundate users with ads or add a feature trying to replicate the success of other platforms (Instagram Reels will never be TikTok!), but everyone’s got to make money somehow.

Some of them will now use Discord to do it.

The post Monetization comes for us all as Discord adds paywalls appeared first on TechHQ.

AI automation in the workplace is moving at a dizzying pace. Business wheels are turning fast to capitalize on the capabilities of large language models (LLMs) such as OpenAI’s GPT-4, which is bringing all kinds of advanced chatbot capabilities to enterprise software thanks to API integration. But LLMs aren’t the only AI automation show in town, and businesses have long been using machine learning methods to enhance sales outcomes – for example, by scaling up authentic content marketing activities.

Authentic content marketing, like other engagement activities, aims to strengthen the bond between a brand and its customers. Potential clients want to know what other users think about products and services. Company websites are great for gathering technical specifications. But it’s impossible for developers to provide an unbiased review of their own products. No firm is going to stay in business long by championing a rival’s offering. And even when a company has the leading solution, and the data to support its claims, customers will likely still want to do more research of their own and seek other opinions.

Don’t just take our word for it

These preferences highlight why firms have been adding authentic content marketing to their sales playbook, and exploring user-generated content (UGC) opportunities more widely. Sources of UGC can take many forms, from social media posts and video unboxings to verified reviews on popular platforms such as Trustpilot, Feefo, and others. The influence of reviews on purchasing decisions, especially for first-time customers, can be huge.

Considering Trustpilot alone, the verified reviews site has compiled over 800 million reviews – accord to figures provided by the firm to TechHQ – with monthly active users totalling over 30 million. And looking at other UGC channels, popular brands receive a multitude of mentions on social media, and this feedback can be gold for firms if they can act on it.

READ NEXT

Responsible sales enablement: let web apps do the heavy lifting

Platforms such as SocialBee help businesses by bringing together company social media accounts into a single pane of glass. Other benefits include analytics and scheduling assistance. Plus, tools are increasingly equipped with AI content generation features, to help write captions and guide other content. StoryChief – another tool that streamlines social media account management for firms – claims that users can work 50x faster using AI.

Time-saving rewards are significant. And so is the potential for companies to use UGC to fuel growth through authentic content marketing. Manual workflows are going to leave too many opportunities on the table. The volume of trusted reviews, social media posts, videos, and other UGC is huge, and it’s a cycle that never stops. But sales and marketing teams can leverage technology to not only keep up, but scale their campaigns in ways that would be impossible by hand. Solutions like Flowbox use AI and automations to gather and organize authentic content posted on UGC channels. Systems can receive content posted by users and compare images and mentions with inventory items in company product catalogs.

Automation allows companies to easily share real-life examples of their products and services being used by customers. And such a portfolio gives potential clients a valuable resource for conducting sales research. Companies wanting to add another layer to their UGC portfolio by working with influencers or brand ambassadors can benefit from AI automation too.

READ NEXT

Open commerce platform takes on big tech

Influencer discovery platforms include affable.ai, which allows firms to search for candidates worldwide across any industry. Companies can quickly generate a shortlist of industry relevant people to engage with thanks to advanced filters. And as relationships form, the system can be used to gather insights on how the influencer-based marketing campaign is going, making it a potential end-to-end solution for firms.

Conventionally, sales enablement is about equipping sales teams with the tools they need to enhance outcomes. But the rise of authentic content marketing and the wealth of business intelligence that goes with it is too valuable to leave siloed. The overlap between sales enablement and content marketing is ripe for cross-functional business teams to explore.

Customers tell the story

The analytics and metrics-based approach at the heart of sales enablement can be readily applied to content marketing. And sales team knowledge of how to keep the conversation going has parallels in the power of customer engagement to boost average order value (AOV). Bazaarvoice – one of the pioneers in helping clients to instill trust and drive up customer confidence through the use of quality UGC – highlights the rewards for firms.

Engagement platforms capable of collecting, displaying, and distributing UGC at scale can dramatically increase AOV. And they point to a growing trend of so-called social commerce where company product stories are told through posts, videos, verified reviews, and other UGC streams. AI and automation helps to bring the content together, and those company product stories contribute across a range of business development channels.

For example, actionable insights can help companies with benchmarking and getting an independent view on their competitors. User feedback can contribute directly to future products and services, as well as improve day-to-day operations. If clients are more confident that a solution will be right for them ahead of purchase, then customer service teams will potentially have fewer product returns or service cancellations to manage.

By enabling authentic content marketing, AI and automation represents a powerful combination for enhancing sales outcomes.

The post Authentic content marketing in an age of AI automation appeared first on TechHQ.

In 2023, it’s crystal clear that large language models (LLMs) are on a trajectory to becoming indispensable enterprise tools. Microsoft is investing billions in its partnership with OpenAI to capitalize on the potential of flagship LMMs such as GPT-4. And the fact that some industry executives are urging for an ‘AI pause’ for six months shines a bright light on the pace of disruption to a huge range of jobs, from managers to models. Adding to that shakeup is an intriguing use of LLMs as silicon personas, which was signposted in a research paper that quietly appeared on the arXiv preprint server in September 2022.

READ NEXT

ChatGPT bug exposes Redis vulnerablity issue

The study, dubbed ‘Out of One, Many: Using Language Models to Simulate Human Samples’, was formally published last month in the journal Political Analysis. But don’t be fooled by the journal’s title; the group’s work could radically change the way that product designers and digital marketers test their ideas and gather insight from potential customers, as we’ll get to.

The research team, which includes corresponding author Lisa P. Argyle, is based in the departments of computer and political science at Brigham Young University in the US. And, realizing that little attention had been paid to the possible applications of LLMs to advance the understanding of human social and political behavior, the group came up with a way of using models – in this case, GPT-3 – as surrogates for human respondents in a variety of social science tasks.

What are silicon personas?

Argyle and her colleagues termed these surrogates, silicon subjects, and these silicon personas – which could equally be created to represent user types – turn out to be incredibly lifelike. In the study, the researchers conditioned GPT-3 with the socio-demographic backstories gathered from 2012, 2016, and 2020 US election survey data, and examined whether they could use their silicon subjects to determine the outcome of the election.

The correlation between the silicon subjects and the actual voter data was incredibly strong, with humans and personas remarkably aligned. And the outcome shows that GPT-3 exhibits sufficiently fine-grained biases and preferences to – in this case – determine which party it would vote for based on the backstory of each of the silicon personas.

READ NEXT

ChatGPT-enhanced Bing mobile app: first impressions

What’s most remarkable about the findings is how capable LLMs are outside of their temporal limits. “The training corpus for GPT-3 ended in 2019, so data from 2020 allows us to explore how the algorithmic fidelity of the language model changes when probed outside the time of the original training corpus,” explained the researchers in their paper.

The persistence of the biases in the model meant that GPT-3 could be used to call the result of the US election in 2020, when presented with silicon subjects conditioned with human-like backstories, even though it lacked training data from that period of time. And if you’re thinking that making a jump from silicon subjects to silicon personas – representing not voters, but customers – could be product testing and digital marketing gold, you could be right.

From digital synapses to digital marketing

But before we start to put flesh on the bones of the next big thing in digital marketing and product testing, let’s take a quick look under the hood of GPT-3. OpenAI’s breakthrough LLM (which was extended and finessed into the basis for the wildly popular ChatGPT) has 175 billion parameters, which was 10 times more than any LLM that had gone before it. The parameters represent the weights at each node in the multi-layered neural network, and determine which of these digital synapses ‘fire’ when GPT-3 is presented with a text prompt and attempts to guess the next word in the sequence.

In principle, the more parameters, the more capable the AI model. But all of those billions of parameters need to be trained, and that commits developers to using vast amounts of data. Initially, parameters carry random values, but over time, during training (which took months, using a custom-designed supercomputer hosted on Microsoft Azure) they converge to their optimum values.

READ NEXT

Will AI take our jobs? Yes and…

Many news articles will claim that GPT-3 is a black box, but you can start to see how LLMs make sense of the world when you consider the sources of the training data. And, contrary to some reports, OpenAI has shared some details on the training dataset that it uses. Training sources include Wikipedia, two internet-based books corpora (equivalent to more than 13 billion words), webpages recommended by upvoted Reddit pages, and a filtered and deduplicated version of Common Crawl – a huge resource containing petabytes of data collected from years of web crawling.

So what does this mean for product testing and digital marketing? Digital marketers have been quick to realize that ChatGPT is useful for rapidly generating copy and applying different writing styles – for example, to create social media posts or other marketing assets. But, as mentioned, silicon personas open the door to much more. And their synthetic data could be used in a variety of user testing scenarios.

Customer insight and product testing smarts

To give you an idea, we ran a super quick experiment using OpenAI’s ChatGPT web UI, feeding the advanced chatbot with the following –

Persona 1 = a middle-aged man who likes sports and has a gym membership.

And then asking ChatGPT to recommend a list of products targeting persona 1, which generated the sequence of suggestions below –

Fitness tracker or smartwatch, athletic shoes, resistance bands, foam roller or massage ball, workout apparel, protein supplements, water bottle, and gym bag.

But the fun doesn’t stop there. For example, you can tell ChatGPT that the year is 1950 and ask the advanced chatbot to regenerate the list, which now includes –

Dumbbells or barbells, athletic shoes, boxing gloves, jump rope, stationary bikes, athletic supporter, sweatbands, and weightlifting belt.

Already, we’ve got some useful insight into product line trends today compared with the 1950s. But this is just scratching the surface. You can go far with silicon personas. But you don’t have to explore this product testing and digital marketing landscape alone. Kwame Ferreira, Hugo Alves, and their colleagues started experimenting with AI as a way of synthesizing virtual users to solve one of their biggest pain points – product testing. And not only did the synthetic data work well, it helped elsewhere too. “As we evolved our experiments in AI it became clearer that we were getting a lot of value in the product discovery phase,” comments the team, which has opened the process as a beta version dubbed ‘Synthetic Users’.

If you have accurate backstories for your clients and customers, there’s a good chance that you’ll be able to put LLM-powered silicon personas to work to deliver user testing data or marketing insight that would otherwise have taken hours, days, or even weeks to collate. And, not only could you save time on your product research, you’ll likely save money too.

The post Silicon personas, ChatGPT & the next big thing in digital marketing appeared first on TechHQ.

Samsung’s camera zoom capabilities are a major selling point for its smartphone range, a one-up against competitor iPhones. But the idea that the phones use a “Space Zoom,” which boasts the ability to take high-resolution shots of objects as far away as the moon, has now been debunked: instead, AI is used to add detail that isn’t actually there to the shots.

Theoretically, the range of phones including the Galaxy S23 Ultra has a 100x zoom level, created by augmenting 3x and 10x telephoto cameras with a digital zoom, aided by Samsung’s AI Super Resolution technology.

Samsung openly uses AI to achieve Space Zoom, assuring that the feature uses up to 20 pictures, then processes them as a composite with AI. The AI identifies the content of the scene and performs a “detail enhancing function” on the subject.

In 2021, claims were made that Samsung’s camera zoom capabilities were fake, and would overlay detail onto any moon-like subject. However, an investigation tried to trigger AI processing onto a clove of garlic and a table tennis ball on a black background to no avail. As such, “proof” that Samsung was faking the moon photos wasn’t found.

Enter, Reddit user ibreakphotos. On Friday, posting to the Android subreddit, the user declared Samsung’s Space Zoom moon photos to be fake. What’s more, they claimed to have empirical proof. They pointed out that no one has successfully shown that Samsung’s camera zoom capabilities are real (or that they’re fake) so far.

According to the post, the user had “always had doubts about [the moon photos’] authenticity, as they appear almost too perfect.” Apparently, the images aren’t “outright fabrications,” but they’re also not completely genuine.

In order to prove the fact, the reddit user downloaded a high-res image of the moon and downsized it to 170x 170 pixels, then applied a Gaussian blur. This removed the detail — “the information is just not there.”

To test the camera, they then put the image full-screen on their monitor and, from the other side of a darkened room, zoomed in on the monitor and took a photo (process available to watch here).

So, the camera zoom is actually just leveraging AI to put craters and other details into places that were really just a blur. This isn’t a case of additional processing, as in super-resolution, when multiple frames are combined to recover detail that would be lost: in this case, a specific AI model, trained on a set of moon images, recognizes the moon and adds a moon texture to the image.

The test proves that there’s no multi-frame sharpening at play, because in that case, every shot would be of the same low-resolution image of the moon. As such, the detail has been added from somewhere other than what the camera was “seeing.”

Harmless enhancement? Samsung’s camera zoom lies

The ethics of AI in photography tie into questions about AI and art. It is worth noting that the experiment relates only to use of the Space Zoom feature to photograph the moon — in other use cases, the zoom capabilities might genuinely just use AI to enhance the images captured (although this user says alarm bells rang when they tried to photograph streetlights a mile away, and they came out blurrier than a picture of the moon).

READ NEXT

AI-enhanced camera latest in “art” debate

As such, it could be that AI is making photography more accessible by allowing hi-res images to be taken from a distance without expensive equipment. However, it could also be argued, as one comment on an appleinsider article says, “you may as well just use Google images.”

What’s raised by the moon photos being fake is the worry that AI is being used without acknowledgement; consumers are engaging with the feature without knowing that it isn’t what it says on the box. Space Zoom is toeing the line between harmless application and ominous misdirection.

Falsifying the detail on a picture of the moon isn’t alarming, but as deepfakes and fraud using computer-generated imagery or text get increasingly common, the issue of transparency around AI needs to be scrutinized.

The post Another moon conspiracy: Samsung’s Space Zoom is actually AI appeared first on TechHQ.

Companies already have a lot to think about from a digital security standpoint. With malware ramping up, phishing and email scams somehow not having been nailed into coffins with stakes through their digital hearts even though it’s 2023, social engineering attacks, business email compromise and the increasing pressure on staff to facilitate insider attacks in a challenging economic environment, really, it’s a lot. And now, according to network security experts Lumu, there’s malvertising.

READ NEXT

Rapid analysis of potential malware threats a reality

There’s what-now?

Malvertising. A combination of… malware and advertising?

Well, yes – but as we found out when we sat down with Lumu CEO Ricardo Villadiego, it’s a little more complicated, a lot more clever, and exponentially more dangerous than you might be thinking.

A history lesson.

THQ:

So – what’s the explanation? What actually is malvertising?

RV:

Well, as you say, partially it’s exactly what you think it is. Back as far as 2000, and even probably before that, the bad actors realized it was easier to trick people into clicking a malware link if there was something trustworthy in front of it – like an ad. To some extent, that’s as old as the hills, right?

THQ:

Fundamental principle of malware to some extent, yes. Click this because it’s real and genuine and something you need to do.

RV:

So they weaponized things like AdWords. The digital networks that existed at the time included things like AdWords from Google, Yahoo ads, and MSN ads. And they paid. That’s the important thing – they paid for the ads, perfectly legitimately – and then added some code that linked the end user to the download of a malware artifact. And that was extremely effective for them, because as a user, when you see an ad that is presented to you from Google, it increases your trust, because you assume it’s been vetted.

READ NEXT

Malware protection: deep learning leads cybersecurity charge

THQ:

The assumption from authority, right? We inherently assume if something’s got an authority’s name on it, or that it’s been allowed through by such an authority, it will have had to meet stringent criteria of safety.

RV:

Right – and we’re more inclined to click on ads that we assume have that in-built safety of authority. That’s how these guys weaponized their online digital advertising channels to deploy malware.

THQ:

We see the point where it’s vital that they bought the ads legitimately. It’s almost like buying an alibi.

Remedial action.

RV:

Yeah. Now, the vendors of the ads began to realize what was happening. So they increased their scrutiny of the campaigns that were paid for, to try and avoid allowing malware through their digital advertisement channel.

The thing about that is that every action creates an adversarial reaction. People love an arms race, and that’s especially true when it comes to cybercriminals and the people who try to stop them.

READ NEXT

Looking For Malware In All The Right Places – Part 2

So what’s happened is that the cybercriminals have seen that the response to their action was tighter controls over the available advertising networks. What do you think they’ve done?

THQ:

Bought Google?

RV:

Not quite.

THQ:

Sorry – gut reaction.

The arms race just got real.

RV:

What they’ve done is create their own advertising networks.

THQ:

O…K. This feels like it’s on us, but we did not see that coming. Digital ad networks… for cybercriminals?

RV:

Yep. Criminal ad networks, hacking perfectly valid websites, and adding JavaScript elements to perfectly legitimate ads, to help them control the ads that that website is going to show to its end users.

THQ:

That’s an almost inspiring level of evil chutzpah.

RV:

And now it creates a serious issue for two main reasons. Number one is that you have tons of websites that are using WordPress, and now they have to take extra steps to secure those WordPress deployments.

And number two, the end user is visiting a valid website, a website that they trust. And that website that they trust is highlighting a piece of art, an image or an advertisement, that is triggering malware and infecting their devices – without the website owners knowing about it. That combination is very powerful because it accelerates the way the bad guys can distribute ransomware precursors that help them to get access to networks, and then execute a ransomware incident.

The end of trust.

THQ:

And that will destroy the trust that people have in the legitimate websites that they like to visit, because all of a sudden those websites are responsible for ransomware, even though the owners and writers and moderators of those sites had no bad intent. It’s anarchy on the internet.

RV:

Absolutely. That was the main reason why the large digital advertisement networks created additional security on campaigns in the first place – they didn’t want the end user to distrust the advertisement network. Now what’s happening is this reaction. The criminals have said “Okay, we’re not going to be able to weaponize a valid advertisement network? We’re going to create our own advertisement network.” If you look at it from a purely objective point of view, it’s really clever how they think.

THQ:

Oh, it’s some Lex Luthor level thinking, right enough.

RV:

They’ve built their own digital advertising network where they can control the ads that people are able to see? And they can resell those capabilities to the ransomware gangs to deploy the initial broker access of their preference over these networks.

THQ:

Malvertising-as-a-service?

RV:

Yep. It is affecting websites today, all across the Americas. And the point is, the potential may not seem like much –

THQ:

The potential seems huge – it can destabilize any trust people have in clicking any ads on any websites.

RV:

Well yes – and it could affect any vertical.

THQ:

Any vertical.

Great.

So, to summarize, cybercriminals developed an effective semi-legitimate way of getting malware onto unsuspecting websites. Authorities tightened up control to try and stop that happening. Annnd now they have their own digital advertising agencies, hiring out the capability to put malware into ads on, let’s say, any WordPress website anywhere, potentially affecting any vertical that exists? Which is more or less to say all verticals, everywhere.

RV:

Pretty much.

THQ:

And that’s malvertising.

In Part 2 of this article, we’ll cover what companies need to do in order to deal with the threat of malvertising.

The post Malvertising – a new corruption appeared first on TechHQ.

Just when you think you can go three days in 2023 without a well-known tech company announcing swingeing layoffs to its workforce as part of a plan to navigate the “uncertain economic climate” of the post-pandemic crunch, you remember that Yahoo still exists.

READ NEXT

Dell embraces layoffs – as a survival strategy

Yahoo has announced it will lose a full 20% of its 8,600 workforce, outstripping the average 6-8% layoffs of many of the bigger players.

Not a Covid readjustment.

While Yahoo has been around since 1994 (making it one of the oldest surviving tech firms of which people have heard), it’s had a challenging time in the last few years, being bought by Verizon in 2017, and then sold to Apollo Global Management in 2021.

Whereas most of the big names in tech are being forced into their layoffs as a result of investment in extra staff to cope with expanded demand during the Covid-19 pandemic and its necessary lockdowns, and those extra staff being unsupportable given an uncertain economic forecast for this year, Yahoo has come up with a different take on its significant layoffs – the first thousand of which are due to take effect within seven days of the announcement.

Rather than cutting jobs across the board (and certainly not on the board), the brunt of the Yahoo layoffs will be felt by the advertising department in its Yahoo For Business arm, which has been particularly struggling in recent years. That department is set to cut more than 50% of its staff across the rest of 2023.

Also, rather than releasing what is becoming an increasingly anodyne repeated meme of regret and sail-trimming, Yahoo is pushing a positive spin on its layoffs. It says the changes will help the company simplify and strengthen its advertising business in the long term, “while enabling Yahoo to deliver better value to our customers and partners.”

READ NEXT

Alphabet and Spotify join the tech layoffs trend

The layoffs are also expected to see Yahoo restructure its advertising focus, to put its “demand-side platform” advertising arm – which has been considerably more successful than that of Yahoo For Business – front and center. That means, for instance, that the occasionally slightly creaky business will be bowing out of the contest for a slice of the digital advertising pie, in which it has had to compete with the likes of Google and Meta.

Like the recently-announced layoffs at Dell Technologies, Yahoo’s layoffs can be thought of as coincidental with the economic bloodbath currently consuming the tech industry, inasmuch as they’re actually not being caused by the same “invest in pandemic, layoff in economic turbulence” strategy of some of the biggest players in the industry.

Smile! You’re out of a job.

But whereas Dell’s layoffs had a distinctly negative connotation (the company actually laid staff off during the pandemic, so the latest layoffs are something of a genuine survival strategy, rather than a simple adjustment which costs thousands of families an income), Yahoo is doubling down on the positive effects of losing 1600 advertising executives and creatives, which it says will see it much more able to compete, not only in the gigantic economic uncertainty of 2023, but also in the future market.

Yahoo’s new overall advertising division will be called – with a clarity of thought which reveals it has been nowhere near an advertising department – Yahoo Advertising. To explain what that will mean for Yahoo’s clients, a company spokesperson said:

READ NEXT

No layoffs at Apple so far — the perks of not overhiring during the pandemic

“In redoubling our efforts on the DSP on an omni-channel basis, we will prioritize support for our top global customers and re-launch dedicated ad sales teams towards Yahoo’s owned and operated properties – including Yahoo Finance, Yahoo News, Yahoo Sports and more.”

The Top Ten.

Of the top ten leading tech companies in the world, most have now laid off a significant portion of their staff in the last few months of 2022 and the first two months of 2023.

• Amazon has announced 18,000 jobs will go.
• Alphabet (owner of Google) will lose 12,000 staff.
• Microsoft has said it will cut 10,000 people loose.
• Tencent will lose around 5,000 staff worldwide.
• Meta has announced it will cull around 10,000 jobs from its infrastructure – but just days ago, was said to be contemplating additional cuts.
• Cisco Systems has axed around 4,800 roles, though like Dell and Yahoo, those losses seem to be part of a broader restructuring effort.
• Oracle is losing thousands of people across 2022/2023.
• And Broadcom has announced almost 2,000 layoffs in the wake of its acquisition by Avago Technologies.

Bucking the trend in the top ten so far, Samsung Technologies – which despite having had what might in most cases be described as a disastrous fourth quarter of 2022, announced it would take on 1,000 new engineers in India to work on its newer technologies, including its AI and machine learning modules.

And then there’s Apple, which with the stubbornness of foresight that makes it both beloved and despised around the world, refused to overhire during the pandemic, and so now has no particular need to tighten its belt – although a combination of late production of its latest iPhones thanks to conditions in its biggest Chinese manufacturing plant, and its domestic labor concerns with unions demanding negotiations and the National Labor Relations Board finding it infringed on its workers’ rights on the way to the founding of its first US unions.

So while it hasn’t been forced into any significant layoffs as yet, Apple’s staffing future may not be as bright as the headline of its pandemic calm may suggest.

Is the wave of layoffs done with the tech industry as we head into the middle of February? It’s too early to tell, but the strong likelihood is that there are more to come, whether, like Dell and Yahoo, as part of fundamental restructuring, or like the majority of the industry, as part of a post-pandemic rebalancing.

Watch this space.

The post Yahoo announces layoffs for 20% of its workforce appeared first on TechHQ.