What Did OpenAI Do This Week? - 07/05/2023 [+30 LINKS]
OPENAI BURNING THROUGH CASH; NO MOAT?
OpenAI is burning through a ton of cash working on things like text to 3D objects, a ton which doubled to $540m in 2022. Previously unreported, the figure reflects the steep costs of training machine-learning models and one reason why OpenAI sold access to ChatGPT earlier than others might have. OpenAI predicted $1 billion in revenue by 2024 but has likely blown passed that already due to popularity and media coverage. All this means costs increase and keep increasing (servers, employees, legal). Sam Altman, OpenAI’s CEO, is reported to have discussed the possibility of a $100B capital raise, with several analysts saying that was both achievable and low. The question this week was if OpenAI (along with Google) has a moat; a competitive advantage that allows a company to maintain its market position and earn outsized profits". People are split. TechCrunch had a unique way of putting it; “Google and OpenAI are Walmarts besieged by fruit stands”. Here’s what the ex-Googler believes that kicked off the whole argument:
Moat or not, OpenAI - and Google - are important parts of the ecosystem. Both Google and OpenAI wouldn’t have been asked to the White House (factsheet) this week if this was the case when other multi-billion user network players (cough: Meta) were not. The US clearly wants to lead, the Biden-Harris administration announced a $140m AI hub investment that will launch seven new National AI Research Institutes, bringing the total to 25 across the country. Altman told reporters after the meeting "we're surprisingly on the same page on what needs to happen." The question remains; where on the spectrum will the hammer come down? Is generative AI moving too fast in open-source circles that the companies who offer some element of control for governments can’t survive or will a larger clampdown come thick and fast across the board? The possibilities on how this ends are murky at best while the technology expands.
SO WHAT?
Right now, any moat is likely to be temporary or attacked regularly thanks to costs, open-source competition and the way AI progresses. What might cause greater concern regarding future revenue guarantees could be regulation, antitrust, brain drain and simply, the ability to innovate against open source models that are faster, more customisable, private, and ‘pound-for-pound more capable’.
WANT TO INNOVATE LIKE OPENAI? Order your copy of the second volume of ‘Disruptive Technologies’ now!
Sam Altman (and other AI powerhouses) met President Biden at the White House to discuss AI dangers. /Reuters
Sam Altman said OpenAI has changed its plans and won’t train on customer data. /CNBC
OpenAI’s contractors unionised in Africa. /Futurism
Sam Altman said AI is a ‘challenge’ that ‘we can manage’. /ABC
Sam Altman and Gregg Brockman gave an interview to Reid Hoffman and Aria Finger. /Possible
Sam Altman shared startup rules OpenAI broke on way to nearly $30B valuation. /Fortune
Sam Altman publicly discusses the remote work debate saying it was a ‘mistake’ and ‘it’s over’ at the Stripe Conference in San Francisco. /Yahoo
OpenAI is working on sharing functionality. /Jane Wong
OpenAI partnered with Box to bring generative AI tools across the platform. /TechCrunch
OpenAI is paying contractors $15 an hour to train ChatGPT. /CNBC
ChatGPT Business will now launch on a dedicated cloud server. /TechRadar
OpenAI partnered with Code.org to launch ‘TeachAI’. /GeekWire
ChatGPT and DALL-E-powered Bing was released to the public. /BI
OpenAI released Shap·E, their latest research in text to 3D. /maginative
OpenAI issued personal data request form. /TechCrunch
Greg Brockman shared information about ChatGPT with Code Interpreter for visualizing any data on Twitter. /Gregg Brockman
OpenAI has 42 open roles it is hiring for, including DALL-E t software engineers. /Sherwin Wu + /OpenAI
OpenAI’s losses doubled to $540 million as it developed ChatGPT. /TheInformation
Sam Altman quoted thoughts on craziness from ‘The Starship and the Canoe’ on Twitter. /Sam Altman
OpenAI will be testing its system against hacking at DEFCON Conference in August. /Neowin
ChatGPT was blamed for a run on EdTech stock results this week. /Reuters
OpenAI may face copyright lawsuits because of how ChatGPT and GPT-4 were trained on copyrighted books. /The Register
Sam Altman said publicly that only major tech growth can fix US debt-to-GDP ratio. /Sam Altman
An ex-OpenAI safety researcher says there's a 50% chance AI could end in 'doom'. /BI
ChatGPT picked 38 stocks and increased its portfolio by 4.9%, outperforming top UK funds. /FT
‘Godfather of AI’ Geoffrey Hinton quits warning of dangers. /Independent
ChatGPT is now banned at Samsung after an internal data leak. /TechCrunch
ChatGPT has been used the create dozens of news content farms. /Bloomberg
Meta issued a ChatGPT scam warning. /DigitalTrends
Slack introduced SlackGPT, Slack but with AI baked in. /Slack
Microsoft outlined how it intends to keep AI ethical. /MicrosoftBlog
OpenAI’s regulatory risks under GDPR, future data scraping and handling “right to be forgotten” requests. /TheVerge