Welcome to The Prompt.
AI is now helping humans correct AI output. OpenAI has trained a new GPT-4-based model called CriticGPT to detect errors in the code produced by ChatGPT. As generative AI models advance and their errors become more subtle, models like CriticGPT could help humans train systems more effectively by detecting more problems. Like other AI systems, CriticGPT isn’t entirely accurate, but humans who use the tool to train AI outperform others by 60%.
Now let’s move on to the headlines.
POLITICS + ELECTIONS
Popular AI chatbots like Microsoft Copilot and OpenAI’s ChatGPT regurgitated disinformation about the presidential debate between President Joe Biden and former President Donald Trump on Thursday, NBC News reported. The responses produced by the Chatbots repeat false statements A conservative journalist said CNN’s broadcast of the debate would be delayed, allowing parts of the footage to be edited before it is released to the public. CNN has denied the allegations.
DATA DILEMMAS
AI research startup Perplexity Increasingly Cites AI-Generated Blogs as Sources and social media posts, which contain inaccurate and contradictory information, Forbes learned. The startup draws data from AI-generated materials on a wide variety of topics spanning travel, politics, sports and medical information. In some cases, the answers generated by Perplexity’s search engine also reflected widespread inconsistencies in AI-generated sources, according to a study by AI content detection platform GPTZero. Perplexity’s chief business officer, Dmitry Shevelenko, said in an emailed statement to Forbes that its system is “not perfect” and that it is continually refining its processes to identify high-quality sources.
In the meantime, Amazon investigates Perplexity to assess whether the startup violated its terms of service by removing websites even after developers tried to stop them, according to CableThe investigation comes after Cable The company discovered that a company-affiliated crawler had ignored a common web standard that allows entities to indicate which websites should not be scraped. Perplexity spokeswoman Sara Platnick said PerplexityBot, the company’s crawler, had complied with the web standard and had not violated the terms of service.
Additionally, Quora’s Poe chatbot allows users to download paid news article files, Cable reported.
TALENT RESHAPING
In order to strengthen its general artificial intelligence team, Amazon has hired the co-founders and some employees of Expert, a startup that builds AI agents that perform various computing tasks. As part of the deal, Amazon will also license two-year-old startup’s technologiesAI models and datasets, the company said. Adept has raised more than $400 million in funding from backers including General Catalyst and Greylock and is valued at more than $1 billion. The deal mirrors a similar partnership entered into by Microsoftwhich in March hired most of Inflection’s employees as well as the startup’s CEO, Mustafa Suleyman, to lead its consumer AI business.
AI CASE OF THE WEEK
Venture capital firm Benchmark raises $425 million for its eleventh fund to support early-stage AI startups, Forbes The firm’s five partners – Peter Fenton, Eric Vishria and Chetan Puttagunta, Sarah Tavel and Victor Lazarte – plan to invest in AI companies in their areas of expertise, such as consumer technology and cloud computing. The investment firm has already backed a number of AI startups, including AI agent startup Sierra and video generation company HeyGen.
DEEP DIVE
Cryptocurrency miners are harnessing their cutting-edge equipment and low-cost energy for the booming artificial intelligence sector.
ILLUSTRATION BY EMILY SCHERER FOR FORBES; GETTY IMAGES
Energy has become THE Artificial intelligence is a hot sector. Take cloud computing provider CoreWeave, for example, which signed a $3.5 billion deal with Core Scientific earlier this month. The deal will see CoreWeave pay $290 million per year over 12 years to host AI-related computing hardware in the Austin-based bitcoin miner’s data centers. CoreWeave will also cover all associated capital expenses.
The deal was so good that Core Scientific’s stock doubled to $10 in early June, leading some observers to consider the company a new “picks and shovels” player for AI. On June 26, CoreWeave announced a second contract for additional infrastructure, which is expected to bring Core Scientific $1.2 billion in revenue in the coming years. Core Scientific emerged from bankruptcy in January and is one of the largest bitcoin miners in North America.
The growing demand for The capacity of the high-capacity computers is fueled by the energy needed for AI applications like ChatGPT: its queries require ten times more electricity than traditional Google searches. That’s an advantage for companies like Core Scientific, which have access to cheap energy in states like Texas and North Dakota. Having enough energy is critical considering that building and connecting new data centers to the grid can take up to six years, according to the research center at Lawrence Berkeley National Laboratory.
“The demand is insatiable,” says Adam Sullivan, CEO of Core Scientific. “If we just implemented what’s in our current power contract, we would be one of the top 10 data center companies in the United States.”
Driven by the rise of AI, data center energy demand could reach 9% of U.S. electricity generation by 2030, according to the Electric Power Research Institute, more than double current consumption.
The shift to AI operations for those with available infrastructure and energy capacity offers potentially compelling benefits. By replacing bitcoin’s volatility with more stable revenue from AI computing, miners can benefit from predictable budgets funded by established customers. It also helps miners increase their revenues to be able to afford the high capital investment needed to remain competitive with new mining equipment, Morgan Stanley analysts concluded in an April report.
Read the full article in Forbes.
YOUR WEEKLY DEMO
NBC Universal will use a AI-generated clone of iconic American sportscaster Al Michaels‘ voice to offer daily recaps Paris 2024 Olympicsthe company said. The synthetic voice, trained on Michael’s past broadcasts on NBC, will be used to deliver personalized coverage of games through the app. Viewers can choose the sports and topics they’re interested in, and the AI-generated voice can render 7 million different versions of recaps, leveraging 5,000 hours of live coverage of the event.
AI INDEX
976
AI-generated news and information sites have been identified by disinformation tracking site NewsGuard.
54
According to the report, AI-generated content farms are aggregated by Google News.
43
Among these AI-generated news sites, some earn revenue from programmatic advertising.
QUIZ
This semiconductor technology company is using light-based chips to help meet AI’s growing power demand for data centers:
- Intel
- Qualcomm
- Samsung
- TSMC
Check if you understood it correctly here.
MODEL BEHAVIOR
Toys ‘R’ Us released an ad made almost entirely with OpenAI’s text-to-video AI tool Sora, to mixed reviews from viewers. (Sora created 80 percent of the ad, and post-production teams edited the video to add the finishing touches.) The ad portrays the company’s late founder, Charles Lazarus, as a young boy who dreams of owning a toy store. People were quick to point out that the the boy’s appearance changes throughout the video And objects continue to melt On social media, people called the advert “crappy”, “hollow” and “like a weird dream”.
