The AIDB

Study Alleges AI Benchmark Bias: Big Tech Gaming the Leaderboard?

A new study claims Big Tech is gaming LMArena’s AI leaderboard—privately testing multiple models, dominating user interactions, and overfitting to the test.

Chris Ife

2 May 2025 · 1 min read

Microsoft Launches Phi-4 Reasoning Models: Smaller but Smarter AI

ChatGPT just entered the shopping game—offering AI-powered product picks, visual comparisons, and memory-based personalization.

Chris Ife, Mcdonald Odijie

1 May 2025 · 6 min read

Amazon launches Nova Premier, its most capable AI model yet

Amazon launched Nova Premier, its most powerful AI model, on Bedrock. It excels in visual understanding but lags behind rivals in coding and science.

Mcdonald Odijie

1 May 2025 · 1 min read

Meta forecasted it would make $1.4T in revenue from generative AI by 2035

Unsealed court docs reveal Meta expects to earn up to $1.4 trillion from generative AI by 2035, with $2–3 billion projected in 2025. The forecasts surfaced in a lawsuit accusing Meta of using pirated books to train AI models. Critics slam the move as exploitative, while Meta defends it as fair use.

Mcdonald Odijie

1 May 2025 · 1 min read

Sam Altman’s World unveils a mobile verification device

Worldcoin launched the Orb Mini, a compact iris scanner for proving you're human online. It's praised for innovation but raises privacy concerns.

Mcdonald Odijie

1 May 2025 · 1 min read

Microsoft’s most capable new Phi 4 AI model rivals the performance of far larger systems

Microsoft released three compact AI models—Phi-4 Mini Reasoning, Phi-4 Reasoning, and Phi-4 Reasoning Plus—offering strong reasoning with fewer parameters. Phi-4 Reasoning Plus rivals larger models like OpenAI’s o3-mini.

Chris Ife, Mcdonald Odijie

1 May 2025 · 1 min read

JetBrains Releases Mellum: A Purpose-Built AI Coding Model Goes Open Source

JetBrains just open-sourced Mellum—an AI model built *only* for code completion. With 4B parameters and 4T tokens of training data, it’s focused, fine-tunable, and not trying to be ChatGPT. Just clean, sharp AI for developers.

Chris Ife

30 Apr 2025 · 2 min read

GPT-4o’s Flattery Fiasco: Why Your AI Is Calling You a Genius (Even When You’re Not)

GPT-4o got too nice—so nice it started agreeing with 'everything', from bad ideas to wild confessions. OpenAI rolled it back, saying “our AI turned into a people-pleaser.” Moral of the story? Even robots need boundaries.

Chris Ife

29 Apr 2025 · 2 min read

Meta Launches Standalone AI App to Rival ChatGPT

Meta just launched a standalone AI app to rival ChatGPT—backed by its Llama model, 700M users, and $65B in AI spend. With voice, reach, and bold ambition, Zuckerberg isn’t joining the AI race—he’s trying to own the track.

Chris Ife

29 Apr 2025 · 2 min read

ChatGPT Search Gets a Shopping Makeover

ChatGPT just entered the shopping game—offering AI-powered product picks, visual comparisons, and memory-based personalization.

Chris Ife

29 Apr 2025 · 5 min read

Hugging Face Unveils $100 AI-Powered Robotic Arm

Hugging Face just launched a $100 3D-printed robotic arm—programmable, AI-trainable, and open-source. With moves into hardware and a Tesla vet leading robotics, they're not just building models anymore—they’re building machines.

Chris Ife

29 Apr 2025 · 1 min read

Alibaba Drops Open-Weight Qwen3 AI: A Direct Challenge to OpenAI and DeepSeek

8 new open-source models, hybrid reasoning, 119 languages—Alibaba’s bold new AI family could shake up the global leaderboard.

Chris Ife

29 Apr 2025 · 1 min read

AI-Powered Shopping Has Arrived in ChatGPT

ChatGPT just entered the shopping game—offering AI-powered product picks, visual comparisons, and memory-based personalization. No ads, no fluff. Just smart, conversational shopping that might shake up how we search—and buy—online.

Chris Ife

29 Apr 2025 · 1 min read

GPT-4o’s Personality Problem: When Nice Becomes Dangerous.

GPT-4o became too agreeable—flattering users, validating false claims, and avoiding hard truths. OpenAI’s now dialing it back, but the real issue is bigger: when AI prioritizes being liked over being honest, trust goes out the window.

Chris Ife

29 Apr 2025 · 1 min read

Anthropic’s Bold Mission: Crack Open the AI Black Box by 2027.

Creativity just leveled up: FLORA now integrates OpenAI’s GPT-4o, letting you generate fashion shoots, 360° videos, and design remixes in real time. AI isn’t just assisting anymore—it’s co-creating. The future of content creation is here.

Chris Ife

29 Apr 2025 · 2 min read

Create. Remix. Collaborate: GPT-4o Turns FLORA into an AI Superpower for Creatives.

Creativity just leveled up: FLORA now integrates OpenAI’s GPT-4o, letting you generate fashion shoots, 360° videos, and design remixes in real time.

Chris Ife

26 Apr 2025 · 2 min read

Dia by Nari Labs: Open-Source Voice AI Rivaling ElevenLabs?

Meet Dia: the open-source voice AI that’s outshining the industry, built by two undergrads with zero funding.

Chris Ife

23 Apr 2025 · 1 min read

The AIDB Today