Category: 未分类

Artificial Intelligence (AI) is revolutionizing numerous sectors, but with the boon comes the bane. AI image generators are becoming more sophisticated, making the task of detecting deepfakes increasingly difficult.

This issue is causing alarm among global leaders and law enforcement agencies who are concerned about the impact of AI-generated deepfakes on social media and in conflict zones.

“We’re getting into an era where we can no longer believe what we see,” says Marko Jak, co-founder and CEO of Secta Labs. “Right now, it’s easier because the deepfakes are not that good yet, and sometimes you can see it’s obvious.”

Jak speculates that we are nearing a point—possibly within a year—where discerning a fake image at first glance will be impossible.

His insights should be taken seriously as he is the CEO of Secta Labs, an AI-image generator company.

The Rising Concerns about Deepfakes

A recent trend in AI-generated deepfakes has sparked outrage and concern. Deepfakes of murder victims have been appearing online, designed to evoke strong emotional reactions and gain clicks and likes.

This alarming trend emphasizes the urgency for more efficient ways to detect deepfakes.

Jak’s Austin-based startup, Secta Labs, which he co-founded in 2023, focuses on creating high-quality AI-generated images.

Secta Labs views its users as the owners of the AI models generated from their data, while the company serves as custodians creating images from these models.

The Call for AI Regulation

The potential misuse of advanced AI models has prompted world leaders to push for immediate action on AI regulation.

This has also led to companies like Meta, the creators of the new AI-generated voice platform Voicebox, deciding against releasing their advanced tools to the public.

“It’s also necessary to strike the right balance between openness with responsibility,” a Meta spokesperson shared.

Deepfakes: A Tool for Misinformation

Earlier this month, the U.S. Federal Bureau of Investigation warned of AI deepfake extortion scams and criminals using photos and videos from social media to create fake content.

In the face of the growing deepfake problem, Jak suggests that the solution may not lie solely in detecting deepfakes, but rather in exposing them.

“AI is the first way you could spot [a deepfake],” Jak said. “There are people building artificial intelligence that you can put an image into like a video, and the AI can tell you if it was generated by AI.”

Technology to Counter Deepfakes

Jak acknowledges that an “AI arms race” is emerging with bad actors creating more sophisticated deepfakes to counter the technology designed to detect them.

Jak proposes that technology such as blockchain and cryptography might offer a solution to the deepfake problem by authenticating an image’s origin.

He also suggests a low-tech solution — harnessing the collective wisdom of internet users.

“A tweet can be misinformation just like a deepfake can be,” he said. Jak believes that social media platforms could benefit from leveraging their communities to verify whether the circulated content is genuine.

As AI advances, the battle against deepfakes continues, underlining the importance of both technological and social solutions to counter this growing issue.

Meta, a leading name in the tech industry, has made a significant leap in artificial intelligence (AI) by developing Voicebox, an advanced tool capable of generating lifelike speech.

Despite the tool’s potential, the company has chosen not to release it immediately due to concerns about potential misuse.

Voicebox

Announced last Friday, Voicebox can create convincing voice dialogue, opening up a range of possibilities, from enhancing communication across languages to delivering lifelike character dialogue in video games.

Unique in its functionality, Voicebox can generate speech it wasn’t specifically trained for.

All it requires is some text input and a small audio clip, which it then uses to create a whole new speech in the voice of the source audio.

In a breakthrough from traditional AI speech tools, Voicebox learns directly from raw audio and its corresponding transcription, eliminating the need for task-specific training with carefully curated datasets.

Moreover, this impressive tool can produce audio in six languages – English, French, German, Spanish, Polish, and Portuguese – offering a realistic representation of natural human speech.

Potential Misuse and Meta’s Precautionary Approach

While Voicebox opens up exciting possibilities, Meta is fully aware of the potential misuse of such a tool.

The AI tool could be misused to create ‘deepfake’ dialogues, replicating the voices of public figures or celebrities in an unethical manner.

To counter this risk, Meta has developed AI classifiers, akin to spam filters, that can differentiate between human speech and speech generated by ‘Voicebox’.

The company is advocating for transparency in AI development, coupled with a firm commitment to responsible use. As part of this commitment, Meta has no current plans to make ‘Voicebox’ publicly available, emphasizing the need to balance openness with responsibility.

Instead of launching a functional tool, Meta is offering audio samples and a research paper to help researchers understand its potential and work towards responsible use.

Global Concerns Over AI Misuse

The rapid advancements in AI are causing concern among global leaders, including the United Nations (UN).

Deepfakes have been utilized in scams and have propagated hate and misinformation online, as highlighted in a recent UN report.

Creating AI tools like ‘Voicebox’ offers numerous possibilities but underscores the importance of cautious development and responsible use to prevent misuse.

As we continue to stride forward in the field of AI, these concerns will remain paramount.

Mercedes is set to revolutionize the way drivers and passengers interact with their cars.

The automaker announced plans to integrate ChatGPT, an advanced conversational AI developed by OpenAI, into its infotainment systems.

The integration is part of a beta program launching from June 16, 2023, giving Mercedes customers in the U.S. a chance to experience more engaging and personalized interactions with their vehicles.

A New Era of Interactive Driving

This innovative development allows Mercedes owners to upgrade their existing MBUX (Mercedes-Benz User Experience) systems with ChatGPT’s functionalities.

A simple voice command, “Hey Mercedes, I want to join the beta program,” will allow the users to enjoy this new addition, enhancing their in-car interactions.

ChatGPT is designed to mimic human-like conversation across diverse subjects. While its capabilities include content synthesis, code writing, and even creative tasks like crafting wedding vows, its role in the car environment remains to be explored fully.

Mercedes believes that ChatGPT’s conversational skills will add value to the driving experience.

“Users will experience a voice assistant that not only accepts natural voice commands but can also conduct conversations,” the automaker stated in a press release.

This feature can provide drivers with comprehensive answers to complex questions, assist with destination details, or even suggest new dinner recipes, all while ensuring their focus remains on the road.

While this integration promises to make car journeys more interesting and engaging, some concerns arise.

Are these wide-ranging functionalities necessary for drivers or passengers? What kind of interactions do users actually prefer while on the move? The answers to these questions are yet to unfold.

Mercedes’ choice to partner with ChatGPT, a third-party service, is a strategic decision to elevate its voice interface service.

However, with this enhancement comes the responsibility of managing user data.

Although the conversations between users and the voice interface are stored in the Mercedes-Benz Intelligent Cloud and anonymized, privacy concerns are still relevant.

Mercedes emphasizes that this data collection is crucial for understanding user behavior, shaping the rollout strategy, and improving the voice assistant across markets and languages.

While the feature’s practicality and privacy implications are subjects of ongoing discussion, one thing is clear: this beta program propels Mercedes towards a future where cars are not just machines, but conversational companions.

Time will tell how this new feature impacts the everyday driving experience of Mercedes owners.

Companies across the globe are increasingly recognizing the value of artificial intelligence (AI) and are willing to pay handsome salaries to professionals proficient in AI tools like ChatGPT.

These companies are offering an average salary of Rs 1.5 crore, with some even offering twice that amount.

Since its launch in 2022, ChatGPT, an AI chatbot developed by OpenAI, has revolutionized the tech industry.

Known for its ability to generate human-like text responses, the bot has found applications in a multitude of areas such as essay writing, music composition, and even poetry crafting.

As a result, expertise in this AI tool has become a hot commodity, opening up several job opportunities.

AI Creating Jobs

According to a study by ResumeBuilder, 91% of companies with job vacancies are seeking candidates skilled in ChatGPT, underscoring the belief that AI has the potential to increase productivity, save time, and enhance overall company performance.

A report by Business Insider indicates that companies listed on LinkedIn are ready to offer annual salaries of up to USD 185,000 (approximately Rs 1.5 crore) to individuals proficient in ChatGPT.

HR company Recruiting from Scratch, based in the US, is currently hiring for the position of Senior Machine Learning Engineer, Audio, with job requirements including familiarity with AI tools and platforms like ChatGPT.

The salary for this role ranges from USD 125,000 to USD 185,000 per year.

Interface.ai, a conversational AI tool, is seeking a Remote Machine Engineer with experience in natural language processing and large language models like ChatGPT, offering a salary up to USD 170,000 per year.

Emergence of Prompt Engineering

Despite concerns about AI leading to job displacement, ChatGPT has actually led to the creation of new job roles.

One such emerging profession is Prompt Engineering, which is growing in popularity.

Earlier this year, San Francisco-based AI startup Anthropic posted a job advertisement for a Prompt Engineer and a Librarian, offering a salary of up to USD 335,000 per year (approximately Rs 2.7 crore).

This demand for AI Prompt Engineering roles extends beyond San Francisco, with numerous job openings for prompt engineers found on platforms like LinkedIn and other job search websites.

Online platforms have also begun to offer courses focused on Prompt Engineering to cater to this growing demand.

Google’s DeepMind, the firm behind the historic victory of artificial intelligence (AI) over human intelligence in the complex board game Go, is gearing up for another breakthrough.

Demis Hassabis, the CEO of DeepMind, has revealed that they are developing a more powerful AI language model named Gemini, intended to surpass OpenAI’s ChatGPT in capabilities.

In 2016, DeepMind’s AI program, AlphaGo, astounded the world by defeating a world champion Go player.

Gemini, The Fusion of Advanced Techniques

Hassabis explained that the company is planning to leverage the techniques used in the creation of AlphaGo for the development of Gemini. Gemini, similar to GPT-4, is a large language model that works with text.

“At a high level you can think of Gemini as combining some of the strengths of AlphaGo-type systems with the amazing language capabilities of the large models.”

Hassabi

The goal is to merge these technologies to equip Gemini with advanced features like problem-solving and planning.

The foundation of AlphaGo’s prowess was reinforcement learning, a method DeepMind has perfected.

The approach involves software learning to solve intricate tasks that require strategic decision-making through trial and error and performance feedback. AlphaGo also utilized a method called tree search to investigate and remember possible game moves.

The intent is to have Gemini imbibe similar methods and push the boundaries of language models, possibly even performing tasks on the internet and computers.

Hassabis indicated that the development of Gemini is a time-intensive process that might take several months and could involve significant funding.

Gemini’s Anticipated Capabilities

The completion of Gemini could significantly bolster Google’s response to the competition from ChatGPT and other AI technologies.

DeepMind was acquired by Google in 2014 after showcasing impressive results from software that employed reinforcement learning to master simple video games.

It demonstrated how AI can perform tasks once believed to be exclusive to human capabilities, often with superior proficiency.

DeepMind’s reinforcement learning expertise could potentially provide Gemini with unique abilities.

However, the journey towards AI advancements is fraught with risks and uncertainties. The rapid development in language models has raised concerns about misuse of technology and control issues.

Despite these concerns, Hassabis emphasized the extraordinary potential benefits of AI in areas such as healthcare and climate science, making it crucial not to halt its development.

Hassabis and his team, while acknowledging the challenges and potential risks, continue to forge ahead, focused on the development of more powerful and innovative AI models like Gemini.

The future of AI seems promising, yet the path ahead remains one of exploration and learning.

Embattled cryptocurrency exchange FTX has abruptly stopped the sale of its stake in AI firm Anthropic, valued over $500 million, as reported by Bloomberg on Tuesday.

FTX had filed for Chapter 11 bankruptcy protection last November, followed by allegations against co-founder Sam Bankman-Fried for money laundering, fraud, and conspiracy to commit wire fraud.

Court documents filed recently alleged that Bankman-Fried and other high-ranking executives at FTX were involved in commingling over $402 million in customer funds.

The collapsed exchange is reported to owe customers around $8.7 billion, with $6.4 billion in the form of fiat currency and stablecoins.

FTX’s Efforts to Repay Creditors

In January, FTX was given the green light by a federal judge to sell off some assets to repay creditors.

Among the assets FTX has offloaded is derivatives trading platform LedgerX, sold for $50 million – a stark loss compared to the $300 million FTX paid for LedgerX in 2021.

Earlier this month, FTX, with the assistance of financial services firm Perella Weinberg Partners, had shown intentions to offload its shares in Anthropic as part of its clawback strategy to repay creditors.

A clawback in bankruptcy is a legal process where a bankruptcy trustee retrieves property or payments made by the company prior to filing for bankruptcy.

Anthropic’s Success Story

Founded in 2021 by former OpenAI employees, Anthropic has become a front-runner in the current AI boom.

The firm launched Claude AI in March after receiving a $400 million investment from Google earlier this year, supplemented by another $450 million in Series C funding in May, led by Spark Capital.

In May, Anthropic made significant advancements in Claude AI, including the development of a rule-set based on the Universal Declaration of Human Rights, designed to promote ethical behavior and discourage undesirable actions.

A Pause on the Sale

The unexpected halt in the sale of Anthropic shares came after several potential buyers had evaluated private information about the stake.

Anthropic, which is privately-held, is currently valued at $4.6 billion, according to Semafor’s June report.

Buyers in the secondary market for shares in private companies have been eager to acquire stakes in Anthropic, even at a premium, indicating the significant potential of this AI startup.

FTX’s stake in Anthropic, amounting to a $500 million investment made by FTX and Alameda, is among FTX’s most coveted assets.

While the reason for this sudden halt remains undisclosed, it marks another twist in FTX’s ongoing legal and financial turmoil.

By Mukund Kapoor

OpenAI, a leading name in the AI industry, has unveiled its new web crawling bot, GPTBot, to broaden the dataset for training future AI systems, possibly including the next version named “GPT-5,” as indicated by a recent trademark application.

Gathering Public Data

The newly released GPTBot will collect publicly accessible data from websites while steering clear of paywalled, sensitive, and prohibited content.

This web crawler functions similarly to those of search engines like Google and Bing, assuming that accessible information is fair for use.

To block the OpenAI web crawler from accessing a site, the owner must add a “disallow” rule to a file on their server.

OpenAI has assured that GPTBot will scan the scraped data to eliminate any personally identifiable information (PII) or text that contradicts its policies.

However, the opt-out approach is generating ethical concerns around consent. Critics argue that OpenAI’s actions might lead to derivative work without proper citation.

Addressing Past Controversies

The launch follows prior criticism where OpenAI was accused of scraping data without permission for training its Large Language Models (LLMs) like ChatGPT.

In response, OpenAI updated its privacy policies in April.

The new web crawler represents OpenAI’s need for more current data to maintain and enhance its LLMs.

The move may indicate a shift from OpenAI’s initial focus on transparency and safety, understandable as ChatGPT remains the most used LLM globally.

OpenAI’s products heavily rely on the quality of data used for training, and the GPTBot aims to gather that essential data.

Competition in the AI Space

Meta, the social media titan, has also been working on AI, offering its model for free unless used by competitors or large businesses.

While OpenAI’s strategy revolves around using crawled data for profitable AI tool ecosystems, Meta aims to build a profitable business around its data.

OpenAI’s ChatGPT currently boasts over 1.5 billion monthly active users, and Microsoft’s $10 billion investment in OpenAI is paying off, as ChatGPT integration has enhanced Bing’s capabilities.

As OpenAI’s GPTBot represents an advancement in AI capabilities, it also reopens copyright, consent, and ethics debates.

As AI systems become more advanced, striking the right balance between transparency, ethics, and technological capability will continue to challenge industry leaders.

The new web crawler’s launch highlights the complexities of innovation in the AI space, where benefits in efficiency and ability may come with potential ethical trade-offs.