This study examined Upper Basic Science teachers’ perceptions and integration practices regarding Artificial Intelligence (AI) technologies in Basic Science teaching in Makurdi, Benue State, Nigeria. A cross-sectional descriptive-correlational survey design was adopted. The population comprised Upper Basic Science teachers in public and private schools in Makurdi, with 47 teachers from 40 selected schools participating. Data were collected using the Artificial Intelligence Perceptions and Integration Practices Questionnaire (AIPI-Q), covering teachers’ perceptions of AI, technologies and purposes of use, and integration practices. The instrument was validated by experts in Science Education and Measurement and Evaluation. Cronbach’s alpha coefficients of 0.91 and 0.86 were obtained for the perception and integration-practice sections, respectively, with an overall coefficient of 0.89. Cohen’s kappa coefficients for the technology-use checklist ranged from 0.76 to 0.92. Data collected were analysed using mean, standard deviations, frequencies, percentages and simple linear regression. Findings showed that teachers had positive perceptions of AI as a useful tool for Basic Science teaching (x̄ = 3.13, SD = 0.75). ChatGPT was the most used AI technology (63%), mainly for information searching (56%), generating lesson ideas (50%), developing instructional materials (47%) and preparing lesson plans (45%). AI integration practices were at the sometimes level (x̄ = 2.65, SD = 1.04). Teachers’ perceptions had a weak positive relationship with integration practices (R = .171) and explained 2.9% of the variance (β = .171, t = 1.166). Thus, positive perceptions alone may not translate into extensive AI integration. Targeted professional development, practical AI competencies, infrastructure and pedagogically purposeful support are needed to strengthen meaningful AI integration in Basic Science teaching.