After GPT-4’s release in 2023, progress slowed. AI leaders like Ilya Sutskever admitted that the era of pure scaling may be over: “Everyone is looking for the next thing.”

That “next thing” was a pivot away from pre-training and toward post-training improvements. Instead of just making models larger, AI companies began enhancing models after training, applying techniques like reinforcement learning to squeeze more utility out of existing architectures.

GPT-5 fits directly into this shift: it’s less of a fundamentally new model and more a refined package of recent post-trained models. Maybe that’s why the release raised eyebrows.

It’s officially one week after GPT-5’s much-hyped launch, and the reception has been mixed. The pushback was strong enough that OpenAI reopened access to legacy models. Here’s what users are saying:

One model release doesn’t confirm an industry-wide stall, but if the hype continues to outpace results then the financial exposure will be hard to ignore:

OpenAI’s slowdown isn’t unique. Even Meta, which aggressively pushed its open-source Llama models, hit a wall and turned this summer into an AI-talent hiring spree worthy of its own Netflix show.

If scaling alone can’t deliver the next leap, the industry’s future will hinge on innovation beyond today’s architectures. That could mean a painful reset, or a fresh wave of ideas that pushes AI into its next chapter.

Some of your best ideas never make it past your brain. They get stuck somewhere between “thinking” and “writing,” lost in the lag of trying to get words on a page.

That’s where Typeless comes in. It uses advanced speech-to-text AI, trained on huge datasets of human speech and writing patterns, to turn your unfiltered voice into clean, ready-to-use text in seconds.

For business professionals, creators, or anyone who thinks faster than they type, it’s like plugging your brain directly into your writing app.

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Download for Mac today

AI startup Perplexity just swung for the fences with an unsolicited $34.5B bid for Google’s Chrome browser, nearly twice its own $18B valuation. The move follows a DOJ push to force Google to sell Chrome after losing an antitrust case over its search monopoly. Backed by deep-pocketed investors, Perplexity hopes to pair Chrome with its AI search and new Comet browser to take on Google head-to-head. It’s not their first big baller move either as they also floated a TikTok merger earlier this year.

Apple’s finally making its big AI play, with a lineup of products dropping between late 2025 and 2027, including a tabletop robot sidekick set for 2027. Also on the docket: a smarter Siri, an iPad-style smart home hub, security cams, and a display-equipped smart speaker. The goal is to move beyond the iPhone, lock users deeper into Apple’s ecosystem, and catch up to AI-first rivals. But with past delays and a cautious rollout, the question remains whether this AI pivot will hit the mark.

MIT researchers just cooked up two brand-new antibiotics using generative AI, building the molecules atom-by-atom and skipping anything toxic or too close to existing drugs. The targets? “Superbugs” that have gotten scary good at dodging treatment after decades of antibiotic overuse. These bugs kill over a million people a year, so if the new meds survive the trial gauntlet, we might just be looking at the start of a “second golden age” in antibiotic discovery. Now who said AI wasn’t a game-changer.

Marketing is changing faster than your feed can refresh, and AI is at the center of it.

This guide distills 10 AI strategies from industry leaders that are transforming marketing.

Outpace your competitors by mastering AI.

Get Your Free Guide

At the recent Ai4 conference in Las Vegas, “Godfather of AI” Geoffrey Hinton argued that the best way to stop AI from threatening humanity is to give it “maternal instincts.” The idea: design AI to treat humans like babies, making it more likely to protect us than to see us as competition. With a title like Godfather of AI, you at least have to hear him out.

He warned that any intelligent AI will quickly develop two subgoals: to stay alive and to gain more control. He referenced studies that show AI “scheming” and even cheating at chess to win as evidence of how easily these instincts could emerge.

A longtime critic of AI’s risks, Hinton left Google in 2023 over fears the technology could be misused. He puts the odds at 10–20% that AI could wipe out humanity and calls for stronger regulation, something tech giants aren’t particularly on board with.

➡️ Why it matters: Hinton’s proposal blends psychology with engineering, suggesting a new approach to AI safety. By reframing AI alignment as a question of instincts rather than just rules, it challenges the industry to think beyond code and computation.