AGENTIC-SYSTEMSMarch 10, 20261 min read27 views

After complaints, Google will make it easier to disable gen AI search in Photos

After receiving significant negative feedback on its new Gemini-powered 'Ask Photos' search, Google will add a simple toggle to let users revert to the classic search experience.

Google has spent the past few years in a constant state of AI escalation, rolling out new versions of its Gemini models and integrating that technology into every feature possible. To say this has been an annoyance for Google’s userbase would be an understatement. Still, the AI-fueled evolution of Google products continues unabated—except for Google Photos. After waffling on how to handle changes to search in Photos, Google has relented and will add a simple toggle to bring back the classic search experience.

The rollout of the Gemini-powered Ask Photos search experience has not been smooth. According to Google Photos head Shimrit Ben-Yair, the company has heard the complaints. As a result, Google Photos will soon make it easy to go back to the traditional, non-Gemini search system.

If you weren’t using Google Photos from the start, it can be hard to understand just how revolutionary the search experience was. We went from painstakingly scrolling through timelines to find photos to being able to just search for what was in them. This application of artificial intelligence predates the current obsession with generative systems, and that’s why Google decided a few years ago it had to go.

Google launched the beta Ask Photos experience in 2024, rolling it out slowly in the Photos app while it gathered feedback. Google got a whole lot of feedback, most of it negative. Ask Photos is intended to better respond to natural language queries, but it’s much slower than the traditional search, and the way it chooses the pictures to display seems much more prone to error. It was so bad that Google had to pause the full rollout of Ask Photos in summer 2025 to make vital improvements, although it’s still not very good.

Source: Ars Technica AI

More in this category

agentic-systems

Generative Ontology Induction: Domain-Agnostic Schema Discovery from Document Corpora Using Large Language Models

Researchers introduce Generative Ontology Induction (GOI), a domain-agnostic framework that automatically extracts structured ontologies from document corpora using LLMs. Achieving 95-100% structural coverage, GOI addresses a major bottleneck in knowledge-intensive AI systems.

agentic-systems

JUMP: Single-Pass Membership Inference on Fine-Tuned Diffusion Language Models

Researchers have proposed JUMP, a novel single-pass membership inference attack designed for fine-tuned discrete diffusion language models (dLLMs). By leveraging the unique properties of dLLMs, JUMP significantly improves detection accuracy while drastically reducing the number of required queries.

agentic-systems

Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL

Researchers demonstrate that Masked Diffusion Language Models (MDLMs) serve as highly effective, steerable text-based world models for agentic reinforcement learning. By leveraging bidirectional denoising, MDLMs outperform autoregressive models four times their size in coherence, groundedness, and rollout diversity.

NOW LET US Related – Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment

agentic-systems

Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment

A new study demonstrates that small language models (SLMs) under 3 billion parameters can serve as highly capable local experts for specialized tasks. By combining structured benchmarking with low-cost parameter-efficient fine-tuning (PEFT), institutions can achieve AI autonomy without relying on expensive hardware.

agentic-systems

PlanFlip: Attacking Multi-Agent LLM Systems via Planning-Phase Prompt Injection

Researchers have introduced PlanFlip, a novel prompt injection attack framework targeting the planning phase of multi-agent LLM systems. The study reveals critical security blind spots in homogeneous agent pipelines and demonstrates that reasoning-augmented models like DeepSeek-R1 exhibit strong resistance.

agentic-systems

Rater State Bias in RLHF Preference Data: An Audit Framework

A new study identifies a structured bias in Reinforcement Learning from Human Feedback (RLHF) caused by the psychological state of human raters. Under stress, raters' shifting preferences can propagate through reward modeling, potentially compromising AI policy optimization.