DIRS.INFO The most relevant business & tech news

☆ ByteDance AI researchers say OpenAI now tries to hide that ChatGPT was trained on J.K. Rowling's copyrighted Harry Potter books
▼ + stars: | 2023-08-15 | by ( Kali Hays | ) www.businessinsider.com time to read: +3 min

Companies like OpenAI are under growing scrutiny over their use of copyrighted material. Prompts about Harry Potter books have showed how common the use of copyrighted work is in LLMs. ChatGPT is trying to hide that it was trained on copyrighted material, according to new research. As do many other AI models, given they've been trained on huge swaths of copyrighted material. The researchers suggested that users who prompt these models to show copyrighted work are "misusing" the technology.

Persons: ChatGPT, Harry Potter, OpenAI's ChatGPT, OpenAI, Microsoft –, J.K, Rowling, Kali Hays Organizations: Google, Microsoft, Tsinghua University China, Twitter Locations: LLMs, Meta, khays

☆ A hedge fund manager shares 2 ChatGPT prompts and the AI plugin he used to filter top stock picks
▼ + stars: | 2023-08-12 | by ( Laila Maidan | ) www.businessinsider.com time to read: +9 min

He fed the LLM data from a terminal and asked it to filter its findings through top academic literature and fund managers. Building on the knowledgeIn his earlier experiments, Patel and his team manually pasted increments of stock data into ChatGPT's text box. Using Julius AI, Patel was able to input data for over 6,800 stocks to see if the outcomes or accuracy would change. The metrics Patel chose to input were based on company fundamentals that he felt were important when picking stocks. Patel concluded that whether it's a human analyst or AI that's picking stocks, there are no guarantees in the stock market.

Persons: Alpesh Patel, it's, Patel, Julius AI, Julius, Price, ChatGPT's, Warren Buffett, Cisco Systems Inc Patel, Johnson, Eli Lilly, Rahul Sonwalkar, you've Organizations: NYSE, Nasdaq, US Securities and Exchange Commission, FTSE, New York Stock Exchange, Alpha, Broadcom, Cisco Systems Inc, Apple Inc UnitedHealth, Johnson, Johnson Exxon Mobil Corp Visa Inc Tesla, Walmart, Taiwan Semiconductor Manufacturing Procter, Gamble, Chevron Corp Mastercard, Depot Inc, Pfizer, AbbVie Inc Merck, Co Inc, Cola Co, Pepsico, Broadcom Inc Alibaba

☆ Deepset, an AI startup that helps companies build apps with LLMs, just raised $30 million with this 12-slide pitch deck
▼ + stars: | 2023-08-09 | by ( Callum Burroughs | ) www.businessinsider.com time to read: +2 min

Deepset, a German AI startup that offers LLM developer tools, has raised $30 million. The company wants to expand its Haystack product and grow its presence in the US. German AI startup Deepset has raised $30 million in a round led by Revolut investor Balderton Capital. Deepset's capital raise comes as AI startups account for a higher proportion of venture capital dealmaking. AI startups raised $25 billion in the opening half of the year against a backdrop of falling VC investment, Crunchbase data shows.

Persons: Milos Rusic, Rusic Organizations: Balderton Capital, Airbus, Siemens, Balderton, GV, Harpoon Ventures Locations: London, Berlin

☆ OpenAI just admitted it has a bot that crawls the web to collect AI training data. If you don't block GPTbot, that's self-sabotage.
▼ + stars: | 2023-08-08 | by ( Alistair Barr | ) www.businessinsider.com time to read: +8 min

Some of these bots have been helpful because they send users to sources of original content online. The most active one is probably Googlebot, which automatically collects web information so Google can later rank and serve it up in Search results. It's called GPTbot and it's being used to scrape and collect online content for AI model training. So what is Clarke's advice for other online content creators when it comes to GPTbot? What is the incentive that OpenAI offers to have these content creators allow GPTbot to crawl and scrape their sites?

Persons: OpenAI, Prasad Dhumal, Neil Clarke, Clarkesworld, Clarke, I've, hasn't Organizations: Morning, Twitter, OpenAI, Associated Press

☆ Inside the secret Google DeepMind mafia of founders working on new stealth AI startups
▼ + stars: | 2023-08-08 | by ( Riddhi Kanetkar | ) www.businessinsider.com time to read: +7 min

There is an influx of cash and interest into artificial intelligence startups right now. Insider spoke to ex-Google DeepMind staffers who have founded AI startups in stealth. He isn't the only DeepMind alum working on practical applications of artificial intelligence. Last month, Mistral, an AI startup founded by DeepMind alum, secured $113 million in seed funding from Lightspeed just four weeks after it launched. He has since been working on his second AI startup in stealth mode since June 2023, adding that working under the radar was inspired by DeepMind's "own model of working in stealth."

Persons: Mustafa Suleyman, Suleyman, Devang Agrawal, Jonathan Godwin, DeepMind, Godwin, Simon Kohl, Ang Li, isn't, Li, Simon Menashy, Adam Liska, Demis Hassabis, GlyphicAI's Agrawal, Mehdi Ghissassi, OpenAI's, Agrawal, Karl Moritz Hermann, DeepMind's Organizations: Google, Microsoft, Nvidia, Labs, MMC Ventures, DeepMind, Lightspeed Locations: DeepMind, London, California

☆ Nvidia reveals new AI chip, says costs of running LLMs will 'drop significantly'
▼ + stars: | 2023-08-08 | by ( Kif Leswing | ) www.cnbc.com time to read: +3 min

Currently, Nvidia dominates the market for AI chips with over 80% market share, according to some estimates. But Nvidia's chips are in short supply as tech giants, cloud providers and startups vie for GPU capacity to develop their own AI models. Nvidia's new chip, the GH200, has the same GPU as the company's current highest-end AI chip, the H100. Oftentimes, the process of working with AI models is split into at least two parts: training and inference. Companies including Google and Amazon are also designing their own custom AI chips for inference.

Persons: Jensen Huang, Walid Berrazeg, Bard, OpenAI's, Huang, Ian Buck, Buck Organizations: Getty, Nvidia, AMD, Google, ARM Locations: Taiwan

☆ Humanity is on the brink of major scientific breakthroughs, but nobody seems to care
▼ + stars: | 2023-08-08 | by ( Hasan Chowdhury | ) www.businessinsider.com time to read: +5 min

Humanity could be on the brink of making major progress in multiple areas of science. These are artificial intelligence, room-temperature superconductors, and nuclear fusion. A lesson from ChatGPT: people get excited by progress when they understand what it means. It's not a surprising attitude from the man who also said "we wanted flying cars, instead we got 140 characters." Efforts remain underway to replicate a room-temperature superconductor's creation.

Persons: Peter Thiel, Thiel, LLMs, chatbot, Marc Andreessen, OpenAI's Sam Altman, Andreessen Organizations: Milken, Peking University, Times, Milken Institute Locations: Korea, Beijing, Silicon, California, ChatGPT

☆ Room-temperature superconductors and nuclear fusion will need a ChatGPT moment
▼ + stars: | 2023-08-08 | by ( Hasan Chowdhury | ) www.businessinsider.com time to read: +5 min

These are artificial intelligence, room-temperature semiconductors, and nuclear fusion. A lesson from ChatGPT: people get excited by progress when they understand what it means. In South Korea last month, researchers declared the discovery of the world's first room-temperature, ambient pressure superconductor – a rock-like material known as LK-99. But if Andreessen and his ilk want the general public to get excited by nuclear energy and superconductors, they have the answer in their own backyard in ChatGPT. Humans need to see and understand the benefits of progress for themselves.

☆ Google reshuffles Assistant unit, lays off some staffers, to 'supercharge' products with A.I.
▼ + stars: | 2023-08-01 | by ( Jennifer Elias | ) www.cnbc.com time to read: +4 min

Google CEO Sundar Pichai speaks on-stage during the Google I/O keynote session at the Google Developers Conference in Mountain View, California, on May 10, 2023. Google wants to "supercharge" its seven-year-old smart assistant using new advancements in generative artificial intelligence, as part of the latest major reorganization of the Assistant unit. A portion of the Assistant team has already started working on the efforts, Ranjan added. As a part of the reorganization, executives announced a dozen changes to the company's "Speech" team, which oversees voice commands. Some members of the Speech team will focus their efforts specifically on Bard, she wrote.

Persons: Sundar Pichai, Peeyush Ranjan, Ranjan, Apple's Siri, Alexa, We've, Axios, Francoise Beaufays, Sissie Hsiao, Bard, Beaufays, Jennifer Rodstrom, we're, It's, Hsiao, OpenAI's, Microsoft Bing Organizations: Google, Google Developers, CNBC, Employees, Microsoft Locations: Mountain View , California

☆ Leaked email shows Amazon formed a new group to work on the 'most ambitious' AI models
▼ + stars: | 2023-07-31 | by ( Eugene Kim | ) www.businessinsider.com time to read: +3 min

Amazon CEO Andy Jassy promoted Rohit Prasad, SVP and head scientist for Alexa, to his direct report. Prasad is now running a newly created AI team at Amazon. Jassy told Amazon's S-team, a group of over two dozen top executives, that he promoted Rohit Prasad, SVP and head scientist for Alexa, as his direct report. In this role, Prasad will lead a newly created group working on the company's "most ambitious" large language models, according to an internal email obtained by Insider. As part of its AI effort, Amazon recently created a new team under AWS focused on helping customers use generative AI, as Insider previously reported.

Persons: Andy Jassy, Rohit Prasad, Prasad, Jassy, Amazon's, Rohit, we've, didn't, Meta's, Dave Limp's, Limp, Daniel Rausch, Eugene Kim Organizations: Amazon, Google, Microsoft, AWS, Alexa

☆ AI researchers say they've found 'virtually unlimited' ways to bypass Bard and ChatGPT's safety rules
▼ + stars: | 2023-07-28 | by ( Beatrice Nolan | ) www.businessinsider.com time to read: +3 min

Researchers say they have found ways to break through guardrails on major AI-powered chatbots. Researchers used jailbreaks developed for open-source systems to target mainstream AI systems. Researchers say they have found potentially unlimited ways to break the safety guardrails on major AI-powered chatbots from OpenAI, Google, and Anthropic. The researchers found they could use jailbreaks they'd developed for open-source systems to target mainstream and closed AI systems. This is something that raises questions about the way AI systems are moderated, as well as the safety of releasing powerful open-source language models to the public.

Persons: jailbreaks, Bard, Anthropic's Claude, they'd, we've, Bard –, Anthropic, OpenAI's, Bing Organizations: Google, Carnegie Mellon University, Center, A.I, OpenAI Locations: OpenAI, Pittsburgh, San Francisco

☆ OpenAI just admitted it can't identify AI-generated text. That's bad for the internet and it could be really bad for AI models.
▼ + stars: | 2023-07-28 | by ( Alistair Barr | ) www.businessinsider.com time to read: +4 min

OpenAI launched a system to spot AI text, but just shut it down because it didn't work. So, developing a way to spot what is AI text versus human text is foundational. In January, it unveiled a "classifier to distinguish between text written by a human and text written by AIs from a variety of providers." Once we can't tell the difference between AI and human text, the world of online information becomes more problematic. I emailed OpenAI to ask about their failed AI text classifier and the implications, including Model Collapse.

Persons: OpenAI, Google Bard, ChatGPT, Biden, Harris, Ilia Shumailov Organizations: Morning, Google, AIs, Bloomberg, Oxford, Twitter Locations: GPT, Cambridge, Toronto

Getting more cars on the road would help Tesla maintain its dominant U.S. market share in "turbulent times" and give it access to precious usage data needed to train the artificial intelligence models behind its self-driving technology. The billionaire believes full self-driving (FSD) could one day account for most of Tesla's value and give it a cushion rivals lack as they try to turn their EV operations profitable. "They're not an AI play the way Microsoft or Nvidia is an AI play," said Thomas Martin, senior portfolio manager at Tesla shareholder Globalt Investments. "They're more of an AI play the way a regular business is an AI play, except that this race to full self-driving has always been an AI issue, and it's always been based on data. Wells Fargo analysts said FSD's adoption could be impeded by its price, which Musk said was low.

Persons: Elon, Musk, Tesla, Gene Munster, Lucy Nicholson, John Tomlinson, Thomas Martin, it's, Wells, Peter Henderson, Aditya Soni, Akash Sriram, Dhanya Ann Thoppil Organizations: Tesla, Deepwater Asset Management, REUTERS, Global, Research, Beta, Microsoft, Nvidia, Globalt Investments, Dojo, Thomson Locations: Los Angeles , California, U.S

[1/2] A man types into a keyboard during the Def Con hacker convention in Las Vegas, Nevada, U.S. on July 29, 2017. In an interview this week, Canadian Centre for Cyber Security Head Sami Khoury said that his agency had seen AI being used "in phishing emails, or crafting emails in a more focused way, in malicious code (and) in misinformation and disinformation." The same month, Britain's National Cyber Security Centre said in a blog post that there was a risk that criminals "might use LLMs to help with cyber attacks beyond their current capabilities." The LLM responded with a three paragraph email asking its target for help with an urgent invoice. Reporting by Raphael Satter in Washington; editing by Chris Sanders and Josie KaoOur Standards: The Thomson Reuters Trust Principles.

Persons: Steve Marcus WASHINGTON, Sami Khoury, Khoury, cybercriminals, Europol, ChatGPT, Raphael Satter, Chris Sanders, Josie Kao Organizations: Def Con, REUTERS, Reuters, cybercriminals, Centre for Cyber Security, European, Cyber Security, Thomson Locations: Las Vegas , Nevada, U.S, Washington

☆ Elon Musk touched on Nvidia, Cybertruck, Warren Buffett, and the dangers of debt on Tesla's latest earnings call. Here's his 12 best quotes.
▼ + stars: | 2023-07-20 | by ( Theron Mohamed | ) www.businessinsider.com time to read: +4 min

Elon Musk mentioned Warren Buffett, Nvidia, cyborgs, and "Game of Thrones" on Tesla's earnings call. Musk is worried about rising credit-card debt and the bewildering economic backdrop. The Twitter owner and Tesla and SpaceX CEO also warned retail investors about the dangers of margin trading, rang the alarm on rising credit-card debt, and said he's totally confused about the state of the global economy. In fact, if you look at the rise in credit-card debt, they're not breaking even every month. Credit-card debt is freaking scary."

Persons: Elon Musk, Warren Buffett, Tesla's, he's, they're, I'm, They've, That's, we've, I've Organizations: Nvidia, Service, Privacy, SpaceX Locations: Wall, Silicon

☆ Apple stock pops on report it's developing its own equivalent of ChatGPT
▼ + stars: | 2023-07-19 | by ( Kif Leswing | ) www.cnbc.com time to read: +1 min

Apple stock popped briefly Wednesday after a Bloomberg News report said the company was developing its own artificial intelligence large language model internally. The move signals Apple is taking recent advances in AI technology seriously and is considering integrating them into future products. For the past year, technologists and investors have been enamored of large language models, an AI technology that can turn out text or code that looks like a human wrote it. Some Apple staffers believe the company is aiming for a significant AI announcement next year, according to the report. Aside from LLMs, Apple uses a lot of machine learning in its products, from Siri speech recognition to the ability of the Photos app to detect faces and pets.

Persons: Siri Organizations: Apple, Bloomberg, Ajax, Microsoft, Google, Meta, Qualcomm

☆ Apple tests generative AI tools to rival OpenAI's ChatGPT - Bloomberg News
▼ + stars: | 2023-07-19 | by ( ) www.reuters.com time to read: +2 min

July 19 (Reuters) - Apple (AAPL.O) is working on artificial intelligence (AI) offerings similar to OpenAI's ChatGPT and Google's Bard, Bloomberg News reported on Wednesday, sending its shares up as much as 2% to a record high. Apple has, however, subtly pushed advanced AI in some of its products such as Apple Photos, on device texting, and the recently launched mixed-reality headset Vision Pro. Apple's core AI product, voice assistant Siri, has also stagnated over the years. The Bloomberg report said several teams are involved in the latest AI effort, which is led by John Giannandrea, the company's head of machine learning and AI, and Craig Federighi, Apple's top software engineering executive. Employees say the tool essentially replicates Bard, ChatGPT and Bing AI, and works as a web application.

Persons: Google's Bard, Apple, Siri, John Giannandrea, Craig Federighi, ChatGPT, Bing, Yuvraj Malik, Devika Organizations: Bloomberg, Microsoft, Nvidia, Apple, Employees, Thomson Locations: Bengaluru

☆ What if AI models like GPT-4 don't automatically improve over time?
▼ + stars: | 2023-07-19 | by ( Alistair Barr | ) www.businessinsider.com time to read: +4 min

GPT-4 users have complained that the OpenAI model is getting 'dumber.' Their findings, published on Tuesday, challenge the assumption that AI models automatically improve. One of the bedrock assumptions of the current artificial intelligence boom is that AI models "learn" and improve over time. This is what users of OpenAI's GPT-4, the world's most-powerful AI model, have been experiencing lately. This recent GPT-4 research paper provides a healthy dose of skepticism to the assumptions that are driving these wild swings in value.

Persons: OpenAI's, OpenAI, Matei Zaharia, Zaharia, Gary Marcus, Marcus Organizations: Twitter, Microsoft Locations: OpenAI

☆ Meta and Qualcomm team up to run big A.I. models on phones
▼ + stars: | 2023-07-18 | by ( Kif Leswing | ) www.cnbc.com time to read: +2 min

But the AI boom has largely missed the companies that make leading edge processors for phones and PCs, like Qualcomm. The announcement on Tuesday suggests that Qualcomm wants to position its processors as well-suited for A.I. If large language models can run on phones instead of in large data centers, it could push down the significant cost of running A.I. Qualcomm will make Meta's open-source Llama 2 models available on Qualcomm devices, which it believes will enable applications like intelligent virtual assistants. Doing this will allow researchers and eventually commercial enterprises to use the AI models on their own computers without asking permission or paying.

Persons: Cristiano Amon, OpenAI's, Bard Organizations: Qualcomm, Milken Institute Global Conference, Calif, Meta, Nvidia Locations: Beverly Hills

☆ This software stock more than doubled on the back of A.I. mania. Here's why Wall Street expects the rally to fade
▼ + stars: | 2023-07-18 | by ( Samantha Subin | ) www.cnbc.com time to read: +4 min

Shares of Palantir have surged 181% this year as investors bet on its potential in the booming artificial intelligence race. The company, founded in 2003, is already making strides in AI, even lending its tools to hospitals . In its latest earnings release in early May, CEO Alex Karp called demand for Palantir's AI platform, which allows commercial and government sectors to use LLMs with their own private data, "without precedent." While some analysts see promise in Palantir's AI capabilities, many remain cautious on the immediate profit contribution of the technology and view the stock as ripe for a pullback. In a note to clients in May, Morgan Stanley's Keith Weiss noted that the rally in Palantir shares and premium valuation creates an unfavorable near-term risk-reward.

Persons: Palantir, Alex Karp, Malik Ahmed Khan, Davidson's Gil Luria, Raymond James, Brian Gesuale, Morgan Stanley's Keith Weiss, Karp, Morningstar's Khan, Mizuho's Matthew Broome, Davidson's Luria, — CNBC's Michael Bloom Organizations: Wall, Nvidia, billings Locations: Denver

☆ Here's 9 ways to turn ChatGPT Plus into your personal data analyst with the new Code Interpreter plug-in
▼ + stars: | 2023-07-15 | by ( Aaron Mok | Gabriel Rivera | ) www.businessinsider.com time to read: +8 min

Here's 9 ways ChatGPT Plus users have used Code Interpreter, from data analysis to game creation. Last week, OpenAI launched a beta version of its plug-in called Code Interpreter to users of ChatGPT Plus, which costs $20 a month. Thanks to the new plug-in, users may now be able to turn ChatGPT into their own personal data analyst. After that, Ker found open source code to help Code Interpreter devise a version of the game. Analyze playlistsWith Code Interpreter, there may be no need to wait all year for your Spotify Wrapped playlist.

Persons: OpenAI, Ethan Mollick, there'd, Mollick, Alex Ker, Ker, Greg Howe, Jason Gulya, Drake Surach, —, Surach, Kris Kashtanova, ChatGPT, Salma Aboukar, Midjourney, Wharton, — Salma Aboukar, Rick Astley Organizations: LinkedIn, ChatGPT, Twitter, University of Pennsylvania's Wharton School of Business, Berkeley College, YouTube

☆ School's out for summer, and if ChatGPT usage is falling because students are off, that's a bad sign for OpenAI
▼ + stars: | 2023-07-14 | by ( Alistair Barr | ) www.businessinsider.com time to read: +4 min

ChatGPT usage recently fell. If that's the reason, it's a bad sign for OpenAI and AI-powered chatbots. ChatGPT usage has fallen. Churn rates, which measure the percentage of users who stop engaging with a service, have also spiked into the 20% range for ChatGPT, according to Bernstein Research. "This idea that if the ChatGPT drop-off is due to students on summer break, that implies a narrower audience and fewer use cases."

Persons: ChatGPT Bernstein, ChatGPT, Mark Shmulik, Bernstein, Shmulik, It's, chatbots, OpenAI, Garry Tan Organizations: Morning, Bernstein Research, ChatGPT Bernstein Research, Microsoft, Enterprise, insider.com Locations: OpenAI, Silicon Valley

☆ Open AI is being investigated by the FTC over data and privacy concerns. It could be ChatGPT's biggest threat yet.
▼ + stars: | 2023-07-13 | by ( Gabriel Rivera | ) www.businessinsider.com time to read: +3 min

The FTC is investigating OpenAI over its lack of transparency regarding data and privacy. The FTC is demanding Open AI detail how and where it collects data. The investigation adds to growing legal challenges filed against the AI company behind ChatGPT. Like the FTC's current probe, many existing legal challenges to OpenAI concern the company's collection of data to train ChatGPT. But Altman has yet to comment on any of the recent legal challenges aimed at his company.

Persons: —, OpenAI, ChatGPT, Sarah Silverman, Sam Altman, Altman Organizations: FTC, Morning, Federal Trade Commission, The Washington, The Post, Twitter, Facebook Locations: Abu Dhabi

☆ FTC investigating ChatGPT-maker OpenAI for possible consumer harm
▼ + stars: | 2023-07-13 | by ( Lauren Feiner | ) www.cnbc.com time to read: +2 min

The Federal Trade Commission is investigating ChatGPT-maker OpenAI to understand if the company has violated consumer protection laws. The Washington Post, which first reported the news, published the FTC's 20-page civil investigative demand (CID), similar to a subpoena outlining key focuses of the probe. A source familiar with the matter confirmed the authenticity of the document to CNBC. The FTC and other agencies have emphasized that they already have legal authority to pursue harm created by AI. The document also asks how OpenAI assesses risk in LLMs and how it monitors and deals with misleading or disparaging statements about people.

Persons: OpenAI, Lina Khan's, Sam Altman Organizations: Federal Trade Commission, Washington Post, CNBC, FTC, CID, ChatGPT, YouTube Locations: Washington, LLMs

☆ An AI detector mislabeled nearly every essay written by a non-native English speaker as being written by a bot
▼ + stars: | 2023-07-13 | by ( Gabriel Rivera | ) www.businessinsider.com time to read: +5 min

Researchers found popular GPT-detectors flagged essays by non-native English speakers as AI-written. Systems that detect AI-generated writing are flagging essays written by non-native English speakers as bot-generated, researchers from Stanford University said. In the study published Monday, the researchers ran more than 100 essays written by non-native English speakers through seven popular GPT detectors. The researchers also fed the detectors essays written by US eighth graders who speak English natively. More than half of the essays written by non-native English speakers were marked as AI-generated by the detection systems, the Stanford researchers found.

Persons: chatbots, James Zou, Zou, OpenAI, Sam Altman, ChatGPT, Altman Organizations: Systems, Stanford University, Stanford, The New York Times

Search resuls for: "LLMs"

25 mentions found