What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
AI can help people make better decisions when its advice is accurate, but people can also lean on incorrect or biased answers. Studies show that risk in particular tasks; they do not establish that more capable models generally make their users worse judges. The key distinction is not simply whether a model improved, but whether its output is reliable for the decision at hand and how a person responds when it is wrong.
Does using AI make you worse at making decisions?
Not necessarily. AI assistance can improve judgment in some experimental settings when the system gives accurate advice. But when recommendations are wrong or biased, people may rely on them too heavily, with worse results. The direction depends on the task, the quality of the output, and how users weigh it—not capability alone.
In a set of experiments involving 1,401 participants, Moshe Glickman and Tali Sharot found that accurate AI improved judgments in their task paradigm, while biased AI amplified human biases in perceptual, emotional, and social judgments. Participants often did not recognize how much the AI had influenced them. The authors describe the study-specific effect this way: “The current study reveals that biased algorithms not only produce biased evaluations, but substantially amplify such biases in human judgements, creating a feedback loop.” Read the Nature Human Behaviour article.
That finding concerns the experiments’ tasks; it is not evidence that AI has already worsened decisions across workplaces, medicine, or civic life.
#1 Best Overall
Why can a faster AI-assisted decision still be a worse one?
Speed, satisfaction, and accuracy are separate outcomes. A CHI 2025 study by Sofia Eleni Spatharioti, David Rothschild, Daniel G. Goldstein, and Jake Hofman compared LLM-based search with traditional search in two online experiments. Participants using LLM search completed tasks faster, used fewer but more complex queries, and reported a more satisfying experience. Accuracy was comparable when the LLM was correct; when it erred, users overrelied on the incorrect information. Microsoft Research summarizes the study.
This is a practical trap: a fluent, convenient answer can reduce the effort needed to reach a decision while also making a mistake easier to accept. The study does not show that all AI use lowers accuracy, or that speed itself causes poor judgment.
Rank #2
How do incorrect recommendations affect human judgment?
People do not always reject bad advice just because it comes from a system they are supposed to review. A 2024 Scientific Reports study used preregistered experiments in personnel-selection tasks and found that incorrect AI advice harmed performance because participants failed to dismiss it. Making the system’s predictions more explainable did not reliably reduce overreliance in those experiments. Read the study in Scientific Reports.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →The lesson is narrow but important: an explanation is something to examine, not proof that a recommendation is correct. A visible rationale may still fail to help a reviewer catch an error.
The OECD’s 2025 report on governing AI uses the term automation bias for overreliance on AI, including accepting incorrect outputs or failing to cross-check them. This is policy synthesis, not a prevalence estimate for everyday AI users. See the OECD report.
Can a second opinion or interface warning solve the problem?
Some design changes have helped in specific studies, but none is established as a universal safeguard.
- Highlighting possible errors: In the second LLM-search experiment, color-coded highlighting helped participants identify errors and improved decision accuracy without changing other outcomes. The study summary is available from Microsoft Research.
- Adding a second opinion: A 2024 ACM study found that presenting another opinion—whether from a peer or another AI—could reduce overreliance while also increasing underreliance. A second voice may alter calibration in either direction; it is not a guaranteed improvement. Read the ACM paper.
- Showing explanations: Explanations did not reliably curb overreliance in the tested personnel-selection experiments, so they should not be treated as a stand-alone fix. Read the study in Scientific Reports.
How can you decide when to trust an AI answer?
For consequential decisions, use AI as input to your judgment rather than as its substitute. These practices follow from the observed risks; the studies do not establish them as a universal intervention.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →- Set your own criteria first. Before reading a recommendation, write down what matters or make an initial judgment. This gives you a point of comparison instead of letting the AI’s framing become your starting point.
- Check important factual claims independently. Look for an authoritative source you can inspect, especially when the answer is confident but offers no checkable evidence.
- Inspect the reasoning without treating it as verification. Ask whether the supporting evidence actually backs the claim. A plausible explanation can accompany an incorrect recommendation.
- Use warnings and second opinions as prompts to check, not as verdicts. They may help you notice errors, but evidence shows they can also shift users toward underreliance.
- Judge performance when the system is wrong as well as when it is right. A model’s ability to solve a task does not by itself show that relying on it improves a person’s decision.
Does a better model mean worse human judgment?
The cited studies do not establish that newer or more capable models cause people’s judgment to deteriorate. They examine different tasks, systems, interfaces, and outcomes; they do not provide one comparable effect size that can summarize “AI makes human judgment worse,” nor do they establish long-term effects of routine AI use on independent judgment.
Best Value
Benchmark reasoning is not the same as human decision-making with AI. A 2024 Communications Psychology paper compares humans and language models on variants of the Cognitive Reflection Test and the Linda/Bill conjunction problem, and discusses how prompts affect model performance. It does not test whether improving deployed models makes their users worse decision-makers. Read the Communications Psychology paper.
To assess an AI-assisted decision process, ask whether the output can be independently verified, how it performs when right versus wrong, whether the interface helps expose errors, whether users make an independent judgment, and whether safeguards reduce overreliance without encouraging underreliance. These are useful comparison questions, not a validated scoring system.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
