Top Tech Trends of 2020: Password Habits, Chrome Boost, AI Advances & Cloud Growth

2020’s Most Common Passwords Remain “123456”

NordPass reports that “123456” topped the 200 most used passwords in 2020, used by 2.5 million people and exposed over 23 million times, followed by “123456789”, “picture1”, and others, with eight of the top ten crackable in under a second.

Chrome 87 Delivers Biggest Performance Gains in Years

Chrome 87’s release brings major under‑the‑hood improvements: active‑tab prioritization cuts CPU use five‑fold and adds 1.25 hours battery life; startup speed up 25 % and page load 7 %; Android Chrome pages load almost instantly; a new PDF viewer and fresh wallpapers are also included.

Gartner Forecasts 18 % Growth in Global Public Cloud Spend for 2021

Gartner predicts public‑cloud end‑user spending will reach $304.9 billion in 2021, a 18.4 % increase from 2020, with SaaS remaining the largest segment and PaaS growing to 26.6 % as remote workers demand scalable, high‑performance infrastructure.

Facebook AI Detects 94.7 % of Removed Hate Speech

Facebook’s AI now identifies 94.7 % of hate‑speech posts that are later removed, up from 80.5 % a year earlier, according to CTO Mike Schroepfer’s blog and the platform’s Community Standards Enforcement Report.

ZTE Introduces AI‑Driven 4G/5G Network Energy‑Saving Navigation

ZTE’s PowerPilot 4G/5G white paper describes a novel AI‑based traffic‑navigation system that steers data to the most efficient network layer, halving energy consumption while preserving user experience.

AMD Instinct MI100 Boosts AI Performance Seven‑Fold

AMD’s new Instinct MI100 accelerator, built on the CDNA architecture, targets high‑performance computing and AI workloads, aiming to push supercomputing toward exascale performance.

Curriculum Learning Improves Natural Language Understanding

By ordering training data from easy to hard, curriculum learning enhances fine‑tuning of pre‑trained models, though defining data difficulty remains a key challenge; the approach shows promise without requiring complex architectures.

EMNLP 2020 Best Paper: Silent Speech Digitization via EMG

Researchers from UC Berkeley use facial EMG sensors to convert silent speech into audible audio, reducing word‑error rates from 64 % to 4 % in one condition and from 88 % to 68 % in another, and they release a new EMG dataset for further study.