OpenAI CEO Sam Altman Shares How Fatherhood Shifted His Perspective

Here is what’s new in the AI world.

Neurochemically Hacked': Sam Altman on How Fatherhood Rewired His Brain

OpenAI’s CEO, Sam Altman, recently described how the birth of his first child had a profound impact on him, one he welcomes wholeheartedly. Altman, whose company created ChatGPT, shared the news of his son’s birth on February 22, noting at the time that both mother and child were doing well and that caring for his newborn was a wonderful experience.

In a recent conversation with Bloomberg’s Emily Chang, Altman reflected on the moment he first met his son. He explained that within an hour of the birth, he experienced what he called a “neurochemical shift” that completely altered his priorities.

Altman acknowledged that this shift is something many new parents go through. He said it happened so quickly that it left him thinking, “Oh, I get to observe this! I’m being neurochemically hacked,” adding that he realized everything in his life was about to change and in a positive way. “It’s the best, most amazing thing ever,” he added.

According to Altman, others have noticed a change in him, too, saying he seems happier since becoming a father. Some even suggest that this new perspective might lead him to make better, more humanity-centered decisions at OpenAI.

Altman married his partner, software engineer Oliver Mulherin, in 2024.

OpenAI appointed Altman as CEO in 2019, the same year he stepped down from his role as president of Y Combinator, according to Forbes.

His entrepreneurial journey includes dropping out of Stanford in 2005 to start Loopt, a social mapping startup he later sold for $43 million in 2012. In 2023, Altman was briefly ousted from OpenAI, only to be reinstated later that year. His estimated net worth stands at $1.7 billion.

Perplexity’s GenAI Turns Rambling Emails Into Pro Spreadsheets

Perplexity, the AI search engine recognized for providing sourced answers to intricate questions, has introduced Perplexity Labs. Currently in beta, this new tool lets users create dynamic web applications, spreadsheets, and dashboards simply by using plain English prompts.

This marks the company's move beyond search into productivity software. Labs allows natural language input to build interactive tools that typically demand coding skills. Whether visualizing stock trends, creating climate trackers, or building live multi-source spreadsheets, Perplexity Labs asserts it handles these tasks and more without any coding required.

From information retrieval to execution

Historically, Perplexity offered a streamlined alternative to traditional search engines by delivering concise, referenced answers rather than link lists. With Labs, the platform empowers users not just to find information but to utilize it actively. Outputs vary from static tables to real-time dashboards that refresh with new data.

For instance, in a demonstration, a prompt to "build a dashboard showing the latest renewable energy adoption rates by country" yielded a polished interface with charts, tables, and verification links in seconds. The company highlights this functionality's value for analysts, journalists, researchers, and leaders managing extensive data.

Entering the competitive AI productivity field

This launch places Perplexity among AI firms developing integrated productivity solutions, competing with offerings like Microsoft Copilot and Google Gemini that enhance document and spreadsheet interactions. Perplexity differentiates Labs by concentrating on generating new tools instead of refining existing ones.

The beta is available to limited users, with a wider release anticipated later in 2025. Perplexity hasn't revealed if Labs will be a paid feature or part of its standard service. Designed for modularity and extensibility, Labs will allow developers to build upon it. APIs and plugins are under development, potentially enabling enterprise clients to embed Labs directly into internal systems and dashboards.

Meta’s Open Source AI Claims Under Scrutiny: What’s the Real Story?

Meta faces renewed accusations of 'open washing' following its sponsorship of a Linux Foundation whitepaper promoting open source AI benefits. The paper emphasizes cost savings for enterprises, asserting that proprietary AI tools cost companies over three times more.

However, Meta's involvement is contentious, as it labels its Llama AI models "open source" despite expert disagreement.

Amanda Brock, CEO of OpenUK, contends that Llama violates accepted open source definitions due to its license restrictions on commercial use. She cites the Open Source Initiative's (OSI) standards, which Llama fails because its commercial limitations breach core open source principles requiring unrestricted use.

Brock recognized Meta's role in fostering open-source AI discussions but cautioned that this mislabelling carries significant risks, especially as policymakers increasingly reference open source in AI regulations.

While Meta consistently markets Llama as open source, the OSI and others persistently counter that its licensing contradicts fundamental open access ideals. This issue isn't isolated; other companies like Databricks faced similar criticism in 2024 over its DBRX model's non-compliance with OSI standards.

The distinction between genuinely open and merely accessible models remains a growing point of conflict within the evolving AI industry.

Bringing Portraits to Life: Tencent Introduces Open-Source HunyuanPortrait

Tencent has recently introduced HunyuanPortrait, a groundbreaking artificial intelligence model engineered to transform static portrait images into dynamic animated videos. This sophisticated model leverages a diffusion-based architecture to synchronize facial movements and spatial positioning from both reference images and guiding videos.

Now open-source, HunyuanPortrait can be accessed through major repositories, but its use is restricted to academic and research applications only.

Open-Source Release by Tencent

On Tuesday, Tencent revealed HunyuanPortrait via its official X (formerly Twitter) account. The model is available on Tencent’s GitHub and Hugging Face pages, and a detailed pre-print outlining its design has been published on arXiv.

The model is strictly for research and academic purposes, and commercial use is prohibited. This open-access approach allows researchers and developers to experiment with HunyuanPortrait’s features and potentially integrate it into their projects.

HunyuanPortrait's Animation Capabilities

The model excels at creating realistic animations from static portraits by leveraging reference videos. It effectively extracts and applies facial cues and head poses from the videos to animate still images.

Tencent claims the model achieves precise movement synchronization and replicates subtle shifts in facial expressions, resulting in highly realistic animations.

Technical Design of HunyuanPortrait

HunyuanPortrait's design draws on the principles of Stable Diffusion models, incorporating a conditional control encoder that separates motion data from identity information in videos.

Motion details are encoded as control signals and then injected into the static portrait using a denoising UNet. This architecture ensures both spatial accuracy and smooth temporal transitions in the generated animation.

Although Tencent asserts that HunyuanPortrait surpasses current open-source models in terms of temporal stability and controllability, these claims are yet to be independently verified. Nonetheless, the model’s architecture positions it as a significant milestone in AI-based animation technology.

Impact on Animation and Filmmaking

HunyuanPortrait holds the potential to transform the animation and filmmaking industries. Traditionally, realistic character animations require either manual keyframing or expensive motion capture systems.

With HunyuanPortrait, animators can simply supply static character designs and desired movements or expressions, and the model will generate high-quality animations efficiently.

By streamlining the animation process, HunyuanPortrait could democratize high-quality animation, enabling smaller studios and independent creators to produce professional-grade content without the need for extensive technical resources or costly equipment. This could spark a surge of creativity and innovation across the industry.4

Ex-iPhone Designer’s $6.4B Success And A Crucial Steve Jobs Lesson

Jony Ive, the former head of Apple’s product design and co-creator of iconic devices such as the iPhone, iPad, and MacBook, recently made headlines by selling his AI device startup, io, to Sam Altman’s OpenAI for $6.5 billion.

Following the sale, Ive reflected on a key life lesson he learned from his long-time collaborator and Apple co-founder, Steve Jobs: the immense power of true focus.

In a conversation quoted by CNBC Make It from a Vanity Fair event in 2014, Ive described Jobs as “the most remarkably focused person I’ve ever met.” He noted that while working with Jobs was challenging, it was also incredibly rewarding.

Ive said, “You can achieve so much when you truly focus,” highlighting that Jobs’ approach to focus continues to surprise him because “so few people actually practise this.”

According to Ive, true focus isn’t something you simply turn on when needed. Instead, it requires consistent dedication, occupying “every minute” of your waking hours. “You don’t decide on Monday, ‘I’m going to be focused,’” he explained.

Ive emphasized that true focus often involves making sacrifices. He recounted how Jobs would frequently ask him, “How many things have you said no to?” Initially, Ive would respond with ideas and opportunities he had passed on, but admitted these were not things he truly cared about, so saying no wasn’t a real sacrifice.

Over time, he came to realize Jobs’ deeper point: “Focus means saying ‘no’ to something that, with every fiber of your being, you believe is a brilliant idea, something you can’t stop thinking about, but you still say no because you’re committed to something else.”

The $6.5 billion acquisition of Ive’s startup, io, marks OpenAI’s largest acquisition to date. This deal secures a dedicated unit for the development of AI-powered devices and ensures the involvement of Ive and his team, many of whom helped design some of Apple’s most iconic products.

This move also represents a high-profile return to consumer technology for the British-born designer, who left Apple in 2019 after years of collaboration with Jobs.

Although Apple CEO Tim Cook hinted at continued collaboration with Ive following his departure, no joint products were released. Now, Ive is embarking on a new partnership with OpenAI’s Sam Altman, whom he described as a “rare visionary.”

Your opinion matters!

Hope you loved reading our piece of newsletter as much as we had fun writing it. 

Share your experience and feedback with us below ‘cause we take your critique very critically. 

Thank you for reading

-Shen & Towards AGI team