Fresh off the wide release this fall of DALL-E, the brilliant AI-powered text-to-image generator, the OpenAI team has done it again.
Six days ago, the AI research team introduced ChatGPT, arguably the most advanced and easy-to-use chatbot to break into the public domain. As more than 1 million users have already discovered, ChatGPT provides rich, intelligent, conversational text in response to complex user prompts.
ChatGPT can write lines of code, write a college-level essay, write responses in the voice of a pirate, and write a Mozart-style piano piece. For more examples of its amazing capabilities, just search for “ChatGPT” on Twitter and prepare to be amazed.
“There is a certain feeling that happens when a new technology adjusts the way you think about computing,” Box CEO Aaron Levie tweeted on Sunday. “Google did it. Firefox did. AWS did it. iPhone did it. OpenAI is doing it with ChatGPT.”
The appetite for ChatGPT has already been so great that OpenAI stopped new subscriptions. A notice on the site Monday morning read: “We are experiencing exceptionally high demand. Please wait while we work to scale our systems.”
ChatGPT is free to use and doesn’t display ads, though OpenAI CEO Sam Altman tweeted on Monday that “we’re going to have to monetize it somehow at some point” due to “exorbitant” operating costs. (OpenAI launched in 2015 with $1 billion in funding from Silicon Valley luminaries, including Elon Musk and Peter Thiel, and received another $1 billion from Microsoft in 2019.)
While ChatGPT remains in its naive incubation phase, its arrival heralds what could become one of the great disruptive events in modern technology.
As some industry insiders and members of the media have pointed out, ChatGPT performs many of Google’s functions, and often does them better than the Alphabet unit. While Google simply gives you the links and tools you need to search for information, ChatGPT can answer elaborate questions, solve intricate problems, and converse in a human way.
How much of a threat could a technology like ChatGPT pose to Google? Considering that Alphabet earned $149 billion in revenue last year from Google Search and other web-based Google properties, the implications are huge.
“The potential for something like OpenAI’s ChatGPT to eventually supplant a search engine like Google is not a new idea, but this release of OpenAI’s underlying technology is the closest approximation yet to how it would actually work on a fully fledged system, and it should scare Google away,” TechCrunch US managing editor Darrell Etherington wrote on Friday.
For now, the idea of AI startups supplanting Google seems premature.
ChatGPT and lesser chatbots still generate incorrect, inconsistent, biased, or dangerously wrong answers at times, undermining trust in the product. (OpenAI willingly acknowledges these issues, calling them a necessary but unfortunate development cost.) Google, by contrast, largely avoids such traps by holding users accountable for examining the information and drawing their own conclusions.
“Search satisfaction is reputation business,” Delip Rao, an artificial intelligence researcher at the University of Pennsylvania, tweeted Saturday. “Once people are disappointed by a search result in something critical, they will use that search interface less and less. For example, Twitter search at https://twitter.com/ sucks, so I use other options to search (sic) my own tweets.” (He’s right. Twitter’s search function is horrible.)
Plus, Alphabet still has plenty of time and money to beat any would-be rival to Google.
Alphabet and its artificial intelligence subsidiary, DeepMind, are heavily invested in the big language models that underpin chatbots, as MIT Technology Review explored in September. And while OpenAI benefits from the massive buzz that comes with the introduction of an early prototype, Alphabet has little to gain by launching early, non-monetized projects that could damage confidence in the company or cannibalize existing revenue.
Even ChatGPT is skeptical about its ability to outperform Google. When The Independent asked ChatGPT if it could replace Google, the bot responded, in part: “It is unlikely that a single search engine, such as ChatGPT, could completely replace Google.”
However, ChatGPT accurately noted that advanced language models offer unique features and a different user experience compared to Google. As the chatbot concluded: “Overall, ChatGPT has the potential to revolutionize the way we search for information online.”
Do you want to send thoughts or suggestions to the data sheet? Drop me a line here.
Falling short in China. Tesla plans to cut production at its Shanghai plant by up to 20% as a result of disappointing demand for the company’s electric vehicles in China, Bloomberg reported Monday. Sources familiar with the decision told Bloomberg that near-term sales forecasts led to the cut, although the Shanghai factory could return to growth if demand picks up. Tesla shares fell 4% in midday trading on Monday.
Getting back to normal. Foxconn believes that its Zhengzhou plant, responsible for assembling most of Apple’s iPhones, will return to full production capacity within a few weeks after COVID-related and labor issues rocked the Chinese facility, Reuters reported on Monday. A Foxconn source told Reuters the company hopes to replace workers who left last month in late December or early January. Foxconn officials announced Monday that company-wide revenue fell 11% year-over-year in November, the period during which employees revolted over COVID-induced shutdowns and broken pay promises.
Waiting for the bill. European Union privacy regulators are expected to issue record fines on Monday against Meta for violating local laws related to the data of millions of users, Politico EU reported. European officials have not commented on the potential scope of the fines, but Meta’s financial statements suggest they could be closer to $2 billion. Irish and French regulators have already fined Meta some $750 million this year for data and privacy breaches.
All good with Apple. Twitter owner Elon Musk said Apple has fully reinstated its ad buying on the platform, the latest sign of easing tensions between the two parties, Bloomberg reported. Musk, who criticized Apple last week for cutting its ad spending on Twitter, revealed the resolution during a Twitter Spaces chat on Saturday. Musk met with Apple CEO Tim Cook after his comment, later tweeting that they had a “good chat” and resolved some misunderstandings.
Best of five. Former WarnerMedia boss Jason Kilar has some ideas about the future of streaming, and they’re not good for fledgling platforms like Peacock and Paramount Plus. In a Wall Street Journal commentary, Kilar predicted that the largely unprofitable streaming industry will consolidate over the next two years, with just five platforms emerging from the crowd. Kilar, who stepped down in April ahead of WarnerMedia’s merger with Discovery, said he believes three platforms will reach the scale needed to generate positive cash flows (he didn’t name names, but Netflix, Disney+ and HBO Max all have chances). -favorites). In the meantime, he suggested that Amazon Prime and Apple+, neither of which he expects their parent company to make big profits, will continue to operate.
From the article:
Despite the chaotic nature of the current moment, a reassuring global constant for decades to come will be our collective need for stories well told. I believe that Hollywood will continue to serve this fundamental human need, but to do so sustainably will first require dramatic change.
There will be multiple commercial casualties in the pay streaming wars and some commercial winners. Digital markets for industries that have high fixed costs and relatively low variable costs have tended to have a few unusually big winners, and I think that will be the case in entertainment.
Slack CEO Stewart Butterfield leaves Salesforce, by Kylie Robison
Elon Musk ‘wanted to punch’ Kanye West after calling the rapper’s swastika tweet ‘incitement to violence’, by Steve Mollman
Bill Ackman denies defending Sam Bankman-Fried, calls FTX fiasco an ‘egregious’ case of ‘gross negligence’ to say the least, by Susanne Barton and Bloomberg
Crypto Billionaire Mike Novogratz Says FTX’s Sam Bankman-Fried Will Go To Jail “If Events Turn Out As I Hope” by Steve Mollman
Meet the power players trying to make FTX clients feel complete, by Jessica Mathews
Circle pulls the plug on SPAC as stablecoin giant announces it’s profitable, by Jeff John Roberts
These 10 employers are still hiring tech and financial talent like crazy, by Sheryl Estrada
BEFORE YOU LEAVE
Exposing age differences. Catfishers might soon have a harder time catching their prey on Facebook. Axios reported Monday that Meta is testing age verification technology on her Facebook dating site, with the goal of replicating the success of similar tools recently implemented on Instagram. Meta officials said some Facebook users looking for love will need to submit video selfies, which the company will feed through software that estimates a person’s age, or upload a picture of their ID. Facebook Dating hasn’t caught on since its launch in 2019, lagging far behind more popular apps like Match Group’s Tinder and Bumble. But Meta officials hope the new steps will provide a greater sense of security and perhaps satisfy pesky European regulators in the process.