Kevin Pereira & Gavin Purcell
160 episodes listed below
PRESS PLAY
Fast Company called AI For Humans 'the most entertaining way to learn about artificial intelligence.' Every Thursday, hosts Kevin Pereira and Gavin Purcell break down the biggest AI news, tools, and breakthroughs, so you can stay ahead without losing your mind. Whether you're AI-curious or deep in the game, this is your weekly reset on everything that matters in artificial intelligence.
Based on 6 episodes we listened to · September 2026
Hands-on tests, sharp industry context and restless humor make fast-moving AI developments easier to understand without sanding off their consequences.
Kevin Pereira and Gavin Purcell turn rapid AI developments into a lively conversation shaped by hands-on experimentation. News about major model releases shares space with tests of AI video, coding tools, robots and agentic AI, while the hosts keep asking what these systems can…
A Sora dog in a Ferrari meets Pentagon AI stakes; you get the joke and the unease.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
GPT-5's launch backlash opens into a grounded look at agents, autonomy, and AI spectacle.
AI For Humans catches GPT-5 at the point where benchmark success stops being the whole story. The episode frames OpenAI's strange week through a launch that delivers state-of-the-art results while still frustrating hardcore users, some of whom push back on the hype and others…
A brisk, funny pass through NVIDIA's AI ambitions, from inference and DLSS 5 to chips in space and a Timothy Chalamet-looking Mario.
NVIDIA's annual GTC turns into a brisk tour of Jensen Huang's push for AI dominance, from infrastructure and chips to platforms and models. Huang's trillion-trillion-dollar ambition sets the scale, while the hosts frame the moment around a claimed shift: AI is now useful enough…
A fast, funny review of AI drug design, NVIDIA money, creative tools, ads, and relationship strain.
AI For Humans turns a week of AI news into a brisk, crowded tour of drug design, data-center money, synthetic media, ads, and domestic awkwardness.
A sharp, spicy AI news round-up about why clickbait often outruns consequential research.
OpenAI's erotica plans give this episode its hook, but AI For Humans treats the headline as a way into a sharper question: why does spectacle move faster than consequence?
A banana-suit image test opens into compute anxiety and an AI cycle fast enough to make March feel dated.
OpenAI's new ChatGPT images lands in the usual pressure cooker: sharper editing and more realistic output instantly put it next to Google's Nano Banana Pro.
A code-red AI race runs into garbled courtroom dialogue and a sharp question about signal in synthetic video.
OpenAI's code red over Google's Gemini 3 Pro gives this episode its live-wire question: has the model race settled, or could rumored OpenAI releases scramble the leaderboard again?
A brisk, funny look at Sora 2, AI video rules, and the remix culture growing around them.
This episode follows the latest Sora 2 nerf and the pushback around it, with the hosts treating the whole thing as a tug of war between fun and guardrails.
Google's AI push anchors the episode, and the rest of the coverage ties its biggest ideas to the wider race without pretending the picture is settled.
The episode stays centered on Google's push in the AI race, with the Made with AI event's voice-edited photos, intent aware apps, and voice to voice real time translation doing most of the heavy lifting.
Markets, macro, AI, and crypto in motion.
OpenAI’s GPT-5.4 arrives as a major new model.
Kevin and Gavin of AI for Humans discuss surging US job cuts exceeding 1 million with AI as a contributing factor, Sam Altman's comments about GPT-6 potentially doing new science, Google's Project Suncatcher sending TPUs to space, and the OpenAI federal backstop controversy.
Google's Project Genie kicks off the conversation.
OpenAI and Anthropic both drop frontier models.
AI is already improving itself.
The hosts of AI for Humans dive deep into Seedance 2.0, ByteDance's new AI video model that produces shockingly realistic clips with celebrity likenesses and voices.
Sora 2 has blown up the internet, and the hosts spend the episode on the weird fallout: cameo clips, fake-life jokes, and the way a new feature turns friends and AI into material for remixes and bad ideas.
OpenAI's automated AI researchers by 2028 pitch sets the frame right away.
Kevin and Gavin open on the newest AI power grab: Apple choosing Google Gemini for its new AI system.
Claude Code is the main event, and the hosts treat it like a software takeover with jokes attached.
AI for Humans covers OpenAI's massive $300 billion Oracle data center deal starting in 2027, exploring what it means about AI's projected growth trajectory and the risk of an AI bubble.
Google’s Nanobanana Pro and Gemini 3 Pro are presented as a real step change, with Kevin Ferrer saying Gemini 3 built the show’s intro as a single-page website and can design websites, code games, and handle video.
Google AI Studio gets the episode's biggest attention: the hosts say it now includes multiplayer experiences, database and authentication help, Firebase support, OAuth continuity, and Next.js integration.
Gemini 3.1, Anthropic’s Sonnet 4.6, Google’s Lyria 3, and China’s Seedance 2.0 anchor a week of AI updates that the hosts frame as “incremental” but fast-moving.
CES 2026 is framed as a robot-heavy week, with NVIDIA's Alpameo autonomous vehicle platform, Boston Dynamics' Atlas, and Unitree humanoids all getting attention.
The show opens with Google looking unusually strong in the AI race.
Nano Banana 2 lands, and the hosts go straight at its better editing and lower price.
The hosts break down Meta's new AI-powered Ray-Ban glasses, highlighting their features and live demo fails, then cover OpenAI's GPT-5 coding advancements, AI video tools on YouTube, and robotics breakthroughs.
AI for Humans dives deep into Google's viral 'nano banana' AI image editing model (Gemini Flash 2.5), demonstrating its remarkable natural language image editing, character consistency, and perspective-shifting capabilities.
AI For Humans opens with OpenAI's surprise GPT-5.1 release, which the hosts describe as a "vibes and safety" update with eight personality types and faster responses.
Salesforce CEO Mark Benioff says the company cut 4,000 jobs because of AI automation, and the hosts connect that announcement to a broader shift in support and ops work.
OpenAI’s GPT-5.2 launches as a new flagship model, and the hosts say it is a step that can beat Gemini 3 on many benchmarks while handling spreadsheet-heavy work better.
The hosts open with the backlash around Sora 2 and the feeling that the fun has been trimmed back.
ChatGPT Atlas is the main event.
Weekly roundup of AI news, including upcoming models from OpenAI and Anthropic, and debates on AI in filmmaking and music.
The hosts discuss rising anti-AI sentiment, new AI models from Anthropic and Meta, and the societal impact of AI.
The hosts discuss Anthropic's new powerful AI model Mythos that is being withheld from the public due to safety concerns, along with OpenAI's new plan and leaked AI image and video models.
Hosts Kevin Pereira and Gavin Purcell discuss the latest AI news, including Anthropic's rapid feature releases and the launch of C-Dance 2.0.
The hosts discuss the release of GPT-5, its capabilities, and other AI news including Google Genie 3 and Eleven Labs' music model.
This episode covers major AI company moves, including OpenAI's competitor to Microsoft Office, Google's coding platform, DeepMind's robotics, and Tesla's RoboTaxi rollout.
The hosts discuss Elon Musk's bid to buy OpenAI, the Paris AI Summit, and new AI video tools from Adobe and Peak Editions.
The hosts discuss OpenAI's new O3 model, its naming, and its performance on the ARC-AGI benchmark, making predictions for AI in 2025.
Weekly AI news show covering big tech controversies, OpenAI's AGI levels, and AI tools with a humorous tone.
The hosts discuss whether AI is overhyped, OpenAI's exit from China, new AI video tools, and the future of hyper-personalized media.