Google Unveils AI Powerhouse Gemini 2.5 Pro Ahead of I/O Fest

Google's release of the Gemini 2.5 Pro Preview, an I/O edition, marks a significant advancement in AI technology, particularly in its ability to transform web application development. This new model introduces improvements that enhance code transformation and editing, making it a versatile tool for developers eager to streamline their workflows. The Gemini 2.5 Pro not only enhances coding efficiency but also boasts a state‑of‑the‑art understanding of video content, reflected in its impressive 84.8% score on the VideoMME benchmark. This dual strength in coding and video comprehension positions the model as a multifaceted resource for developers and businesses alike, promising increased productivity and innovative application potential. It is accessible through platforms like the Gemini API, Vertex AI, AI Studio, and the Gemini chatbot app, maintaining the previous version's pricing, thereby providing a cost‑effective solution to its users.

The anticipation surrounding Google's Gemini 2.5 Pro Preview was palpable as it was released just before the Google I/O Developer Conference. This updated AI model is expected to highlight Google's capabilities and herald further advancements they plan to showcase at the event. By improving its coding capabilities significantly, particularly for web app development, code transformation, and editing, Google addresses a growing demand for sophisticated AI tools that enable developers to efficiently build complex applications. The I/O edition is specifically designed to offer distinct enhancements over its predecessor, including a reduction in tool‑calling errors and improved performance in tasks such as video understanding, thus directly responding to developer feedback. Moreover, the continued focus on improving AI integrations in everyday applications through platforms like Gemini reinforces Google's commitment to embedding AI functionality into various digital environments, fostering a seamless user experience.

Public reception of the Gemini 2.5 Pro Preview (I/O edition) has been overwhelmingly positive among developers and technology enthusiasts. Many have praised its improved coding capabilities, which are crucial for tasks that require high reliability and efficiency, such as building interactive web applications. There is also excitement about its performance in video understanding, which has various applications, including content moderation and automated video editing. However, despite this enthusiasm, there are challenges that Google must address. Reports of 'Internal Error 500' and other performance issues have surfaced, particularly on iOS devices, and on the Google AI Developers Forum, some users have expressed concerns about rate limiting and inconsistent outputs. Addressing these issues will be vital in ensuring reliable functionality and maintaining user trust in the model's capabilities.

Google's launch of the Gemini 2.5 Pro Preview (I/O edition) introduces significantly enhanced coding capabilities, making it a game‑changer for developers looking to create interactive web applications. This updated AI model shines in areas such as code transformation and editing, providing an invaluable resource for developers aiming to streamline and enhance the efficiency of their coding projects. The Gemini 2.5 Pro is available through Google's comprehensive suite of AI tools, including the Gemini API, Vertex AI, and the AI Studio, maintaining the pricing structure of its predecessor, ensuring accessible innovation for its users.

A hallmark of the Gemini 2.5 Pro's enhanced capabilities lies in its ability to bridge the gap between developers and robust software architecture, particularly in building interactive and complex web applications. The AI's refinement in code transformation and editing not only accelerates the development process but also enhances the overall quality of software output. Its enhanced functionality supports developers by swiftly converting conceptual ideas into practical applications, thus accelerating innovation in the tech industry. Furthermore, its state‑of‑the‑art video understanding capabilities, highlighted by its impressive score on the VideoMME benchmark, augment its proficiency beyond traditional coding tasks.

The strategic release of Gemini 2.5 Pro Preview aligns with Google's ambitions to demonstrate its prowess at the I/O developer conference, showcasing advancements that position it at the vanguard of AI innovation. By offering sophisticated tools for web app development and video understanding, this edition exemplifies Google's response to growing demands for versatile AI solutions that can adapt to a multitude of development needs. The inclusion of improvements in function calls and error reduction further enhances its utility and user experience, reflecting the culmination of developer feedback and technological foresight.

The launch of Gemini 2.5 Pro has caught significant attention in both technical and commercial spheres. Available through multiple platforms such as the Gemini API, Vertex AI, AI Studio, and the Gemini chatbot app, Gemini 2.5 Pro ensures widespread accessibility for developers seeking to leverage its advanced capabilities. The model's pricing is designed to be consistent with its predecessor, enabling users without budgetary surprises. This strategic approach facilitates ease of transition for current users while attracting new developers intrigued by the model's enhanced features.

By maintaining the same pricing structure as the previous version, Google has effectively positioned Gemini 2.5 Pro as an attractive offer amidst competitive AI models on the market. The pricing consistency underscores Google's confidence in the model's advanced features such as improved coding and video understanding capabilities while removing financial barriers for users eager to engage with the latest in AI technology.

Accessibility and affordability are central to Gemini 2.5 Pro's market strategy. The consistency in pricing ensures that cost does not become a deterrent for developers enthusiastic about integrating the latest AI advancements into their projects. Furthermore, the model's availability on prominent Google platforms ensures seamless integration into existing workflows, reinforcing its position as a valuable resource for various application domains.

Gemini 2.5 Pro's pricing strategy reflects a commitment to accessibility, harmonizing advanced AI capabilities with a sustainable cost model. This approach not only solidifies its presence in the AI ecosystem but also encourages broader adoption, aligning with Google's vision of embedding intelligent solutions into diverse technological applications. As such, Gemini 2.5 Pro is not just poised to enhance productivity but also to democratize access to cutting‑edge AI development tools.

Google's debut of the Gemini 2.5 Pro Preview, termed the I/O edition, marks a significant milestone in AI development, reflecting both technological progress and strategic timing. The designation "I/O edition" highlights the model's unveiling just ahead of Google's annual developer conference, I/O, which serves as a platform for major product announcements and technological innovations. This strategic release signals Google's intention to emphasize its latest AI advancements during the conference, potentially setting the stage for competitive developments in the tech industry. The I/O edition not only boasts improved technical capabilities but also represents Google's commitment to showcasing its forefront position in the fast‑paced landscape of artificial intelligence .

The significance of the "I/O edition" extends beyond improved functionalities; it embodies Google's adaptive strategies in responding to market dynamics and developer feedback. With capabilities such as enhanced web app development, code transformation, and video understanding, the Gemini 2.5 Pro Preview addresses critical areas of developer needs. This release is not just about showcasing an updated model but also about solidifying Google's position in the increasingly competitive AI development market. By aligning the launch with the I/O conference, Google leverages its global platform to maximize visibility and impact, thereby fostering anticipation and strengthening its market presence. The anticipatory timing of the I/O edition reflects a broader trend of major tech companies synchronizing product enhancements with key industry events to optimize both technological narrative and brand momentum .

The "I/O edition" of Gemini 2.5 Pro exemplifies a blend of technological advancement and strategic foresight. By choosing to unveil this model prior to Google I/O, the company underscores its strategy of using its developer conference as a launchpad for the most cutting‑edge technology. This new edition is not just an iterative improvement but a deliberate effort to set new standards in AI capabilities, particularly in areas like coding and video understanding, thereby appealing directly to developers and industry stakeholders. Such strategic timing is symptomatic of Google's larger organizational goals to influence industry standards and solidify its leadership role within the AI domain. Consequently, the I/O edition acts as both a product innovation and a strategic maneuver to sustain Google's trajectory of growth and influence in the global technology landscape .

The performance of AI models in video understanding tasks is a significant indicator of their advanced capabilities. Google’s latest Gemini 2.5 Pro Preview (I/O edition) demonstrates remarkable proficiency in this area, securing a top‑tier score of 84.8% on the VideoMME benchmark. This achievement highlights the model's comprehensive ability to interpret and process video content, showcasing the strides made in AI's cognitive processing capabilities. As video content becomes increasingly prevalent, the demand for models that can effectively analyze and comprehend video data escalates, making Gemini 2.5 Pro’s performance an indicator of its cutting‑edge technology.

The robust performance in video understanding tasks by Gemini 2.5 Pro Preview (I/O edition) is not just about achieving high scores on benchmarks like VideoMME; it also signifies a leap forward in practical applications for businesses and developers. This capability can enhance various sectors such as media, entertainment, education, and security by automating tasks like video analysis, content moderation, and real‑time video monitoring. Consequently, Google's continual efforts in refining AI capabilities are setting a high standard in the AI industry, contributing to more efficient workflows and innovative solutions in handling video content. This positions the model as a valuable tool for industries reliant on video data analysis.

Moreover, the advancements in video understanding align with Google's overall strategy to integrate AI into everyday applications, reflecting a broader trend within the tech industry. The ability to understand and process video content efficiently ensures that developers and businesses can harness AI's full potential, optimizing processes and uncovering new opportunities in digital ecosystems. By making such AI capabilities widely accessible through platforms like the Gemini API, Google is contributing to a more interconnected and AI‑driven future, enabling organizations to leverage sophisticated AI tools without extensive technical know‑how. This growth trajectory promises to further transform how video content is perceived, analyzed, and utilized in various domains.

Public and developer reactions to Google's release of the Gemini 2.5 Pro Preview (I/O edition) have been a mix of enthusiasm and critique, indicative of the broader perceptions that often accompany significant technological advancements. The developer community, in particular, is energized by the enhanced coding capabilities that this model offers. These improvements are seen as a significant leap forward, especially in terms of building interactive web apps, transforming code, and editing. This sentiment is echoed by those who have witnessed the model's improved performance rating on platforms like the WebDev Arena Leaderboard. Such achievements have cast a positive light on Google's efforts to meet and exceed the expectations of its user base. The introduction ahead of the I/O conference has only further amplified interest, with the anticipation of more revelations at the event itself. More about its features can be found in the detailed coverage by TechCrunch.

However, this optimism comes with a fair share of challenges. While the model's positive aspects continue to attract attention, some early adopters have voiced concerns regarding operational hiccups within the Google AI Developers Forum. Feedback highlights issues such as the 'Internal Error 500' and 'google.api_core.exceptions.Cancelled: 499' errors, which have caused disruptions for users who rely on stable performance for development activities. Additionally, reports of slow file uploads and inconsistent performance on iOS devices have surfaced, prompting calls for quick resolutions from Google's technical teams. Despite these setbacks, there's a prevailing belief within the community that Google's commitment to addressing these issues through subsequent updates will bolster confidence in the long‑term stability and usability of the Gemini 2.5 Pro Preview. Insight into Google's approach to handling these issues is discussed in threads available here.

Moreover, the public reaction to the model is shaped by the ongoing narrative of AI integration into everyday applications and development tools. Users are particularly impressed by how the model not only advances current capabilities but also aligns with broader trends in AI, such as improved function calling and video understanding applications. This alignment with current tech trends reinforces the perception of Google as a leader in AI innovation. Yet, the shadow of potential issues like rate limiting and character encoding errors serves as a reminder of the complexities involved in pushing technological boundaries. These mixed reactions provide a comprehensive overview of the current landscape facing Google's latest AI offering, with the potential for these issues to be rectified, fostering even greater enthusiasm.

Experts are buzzing about Google's Gemini 2.5 Pro, highlighting its potential to revolutionize AI applicability in various domains. Michele Catasta, President of Replit, appreciates the model's remarkable capability over latency ratio, which translates to effective handling of tasks that demand high reliability. This emphasizes its ability to support developers without compromising speed, ushering in more efficient coding processes [TechCrunch](https://techcrunch.com/2025/05/06/google‑debuts‑an‑updated‑gemini‑2‑5‑pro‑ai‑model‑ahead‑of‑i‑o/).

Silas Alberti from Cognition has noted in particular the model's acumen in handling complex software refactoring tasks, like the optimization of request routing for efficient backend operations. This capability showcases Gemini 2.5 Pro's profound understanding of software architectures, a crucial element in evolving AI's role within enterprise applications [TechCrunch](https://techcrunch.com/2025/05/06/google‑debuts‑an‑updated‑gemini‑2‑5‑pro‑ai‑model‑ahead‑of‑i‑o/). Such enhancements underscore the model's readiness to meet professional demands in coding and software development environments.

Michael Truell, CEO of Anysphere Inc., emphasizes Gemini 2.5 Pro's improved performance within Cursor, an AI‑powered coding editor. The significant reduction in tool‑calling errors and an increase in trigger rates signal a leap forward in usability. These advancements not only make the tool more user‑friendly but also highlight Google's commitment to improving AI accessibility and effectiveness in practical scenarios [TechCrunch](https://techcrunch.com/2025/05/06/google‑debuts‑an‑updated‑gemini‑2‑5‑pro‑ai‑model‑ahead‑of‑i‑o/). Such feedback from industry leaders elucidates Gemini's pivotal role in the future of AI‑driven development tools.

In launching the Gemini 2.5 Pro Preview (I/O edition), Google has made noteworthy strides in catering to developer feedback, yet certain initial user concerns demand attention. Chief among these are reports of persistent 'Internal Error 500' and 'google.api_core.exceptions.Cancelled: 499' errors, as noted on the Google AI Developers Forum. These errors can potentially disrupt workflow and diminish user experience, indicating that Google's engineering teams will need to prioritize bug fixes to strengthen the model's stability.

Performance‑related complaints have surfaced, particularly around the speed of file uploads and the functionality of the Gemini Live API on iOS platforms. Such issues might impede the model's utility in fast‑paced development environments where quick uploads and real‑time API responses are crucial. Thus, users are looking forward to forthcoming updates where these technical hitches can be resolved, in order to fully capitalize on the AI's advanced capabilities.

In addition to technical errors, users have reported inconsistent outputs, such as challenges with handwritten text recognition and character encoding, which indicate potential gaps in the system's robustness. Overcoming these initial setbacks is vital for ensuring that the Gemini 2.5 Pro Preview (I/O edition) becomes a reliable tool for developers. Continual refinement and adjustments will likely be part of Google's strategy to address these teething problems, thereby enhancing user satisfaction and trust.

Rate limiting has also emerged as a significant constraint, affecting developers who require uninterrupted access to the model for extensive testing and application. With user reports illustrating a drop in rate limits, feedback from the community is likely being used to inform further calibrations of service tiers and accessibility to balance demand with system capacity. Addressing these limitations will be crucial in supporting a scalable user base and ensuring inclusive access to the model's functionalities.

Overall, early adopters of the Gemini 2.5 Pro Preview are optimistic yet cautious. While the enhanced capabilities in coding and video understanding have generated excitement, the presence of issues cannot be overlooked. Google's proactive engagement with developers and timely updates will be key in resolving these concerns and securing the confidence of its user community. As these updates are rolled out and issues resolved, satisfaction and widespread adoption are likely to grow.

The introduction of Google's Gemini 2.5 Pro Preview (I/O edition) represents a significant milestone in AI technology with far‑reaching implications. With enhanced coding capabilities, this model promises to revolutionize web app development, code transformation, and editing. Its availability through platforms like the Gemini API, Vertex AI, and the Gemini chatbot app facilitates widespread accessibility, allowing more developers to benefit from its advanced functionalities. By improving coding efficacy, businesses stand to gain substantially through increased productivity and reduced operational costs, potentially sparking innovation across various sectors. Particularly notable are the model's state‑of‑the‑art features in video understanding, achieving remarkable scores on benchmarks like VideoMME, which suggest new applications in fields ranging from content creation to automated editing.

The economic impact of AI enhancements, such as those introduced with Gemini 2.5 Pro I/O edition, can be profound. For enterprises, improved coding capabilities mean faster development cycles and streamlined workflows, translating into substantial cost savings and profit maximization. The improved video understanding could revolutionize industries reliant on media analysis or content moderation. While these developments are promising, the model's early adoption is marred by user‑reported issues including functional errors and performance challenges. These setbacks must be addressed to ensure the model's reliability and cost‑effectiveness, vital for businesses seeking to integrate AI into their operations efficiently.

Socially, the advancements in AI as demonstrated by Gemini 2.5 Pro hold both promise and peril. Improved capabilities in video understanding offer opportunities for enhanced educational tools and communicative interfaces, fostering better learning and sharing experiences. However, these technologies also raise ethical concerns, especially regarding the generation of deepfakes and the spread of misinformation. This dual potential underscores the necessity for responsible AI deployment, with safeguards against misuse and robust frameworks to ensure ethical usage.

From a political standpoint, the concentration of advanced AI capabilities such as those within Gemini 2.5 Pro by major corporations like Google prompts essential discussions on data privacy and algorithmic fairness. As AI's influence expands, it becomes imperative to craft policies that address concerns around AI‑driven surveillance and bias. The global nature of these technologies calls for international collaboration in AI governance to navigate these challenges effectively, ensuring that AI serves the public good without compromising individual rights and freedoms.

Overall, while Google's Gemini 2.5 Pro Preview I/O edition is poised to set new standards in AI capabilities, its future implications are multifaceted and complex. The model must overcome early performance challenges to achieve widespread adoption. Moreover, its long‑term impact will depend on the establishment of responsible development practices and regulatory frameworks that govern its use. By doing so, it ensures that such powerful technologies contribute positively across economic, social, and political arenas. The early enthusiasm surrounding its launch must be matched with continuous support and refinement to realize its full potential.

The launch of the Gemini 2.5 Pro Preview (I/O edition) by Google marks a significant milestone in artificial intelligence, poised to have profound economic implications. One of the most notable impacts is on the software development industry, where the enhanced coding capabilities of Gemini 2.5 Pro promise to streamline development processes and reduce costs. By improving efficiencies in web app development, code transformation, and editing, businesses can anticipate faster development cycles and reduced labor expenses. This not only increases profits but also enables them to focus more resources on innovation, potentially fostering greater technological advancement across various industries. As a result, companies that integrate this AI model into their workflow may gain a competitive edge in terms of productivity and creativity, ultimately driving the digital economy forward. Furthermore, the availability of Gemini 2.5 Pro through various platforms such as the Gemini API, Vertex AI, and AI Studio broadens its accessibility, thereby democratizing advanced AI technology. This broader access is likely to catalyze a surge of innovations as startups and smaller businesses harness the power of AI to solve complex problems. The economic impact of such democratization can be significant, as it levels the playing field, allowing smaller enterprises to compete with established players by utilizing cutting‑edge AI technologies. This could lead to a more dynamic and competitive market environment, resulting in a proliferation of new business models and opportunities. However, while Gemini 2.5 Pro presents promising advancements, issues such as early reports of errors and performance glitches must be addressed to ensure reliability and maximize its economic potential . Experts emphasize that the efficiency and cost‑effectiveness brought by Gemini 2.5 Pro's capabilities depend heavily on its operational stability. Leaders in the tech community like Michele Catasta and Silas Alberti highlight the potential for significant productivity gains and enhancements in enterprise applications . These improvements in AI models translate to tangible economic benefits, such as reduced operational costs and higher margins for businesses. However, the adoption of Gemini 2.5 Pro may face hurdles if technical issues persist, affecting its overall cost‑effectiveness and reliability in real‑world applications. Google's commitment to resolving these challenges will be crucial to its broader acceptance and success in the market, where consumer trust and satisfaction are paramount.

The release of Google's Gemini 2.5 Pro Preview (I/O edition) has sparked discussions around its social implications and potential ethical dilemmas. By enhancing capabilities in video understanding, this AI model opens new possibilities in education, particularly in creating interactive and engaging educational tools. However, the sophistication in processing video content could also lead to challenges, including the creation and spread of deepfake videos. Such technology could fuel misinformation campaigns, posing risks to societal trust and the integrity of information dissemination. Google must prioritize ethical considerations in further development to counter these risks and ensure responsible use of such powerful tools.

Moreover, the integration of Gemini 2.5 Pro into everyday applications signifies a shift towards more ubiquitous AI presence in our daily lives. While this can enhance productivity and convenience, there are ethical considerations around data privacy, consent, and the potential for algorithmic bias. As AI becomes embedded into more platforms accessible to the wider public, users' data could be handled without explicit consent, raising privacy concerns. Google's responsibility to maintain transparent and ethical data management practices becomes paramount in maintaining public trust.

Furthermore, ethical considerations extend to the accessibility and inclusivity of AI advancements. As AI models like Gemini 2.5 Pro become more integral to technology and business processes, ensuring that these innovations do not exacerbate social inequalities is vital. The digital divide could widen if such technologies are only accessible or beneficial to larger entities with significant resources, leaving smaller businesses or less tech‑savvy populations at a disadvantage. Google must strive to create equitable solutions that democratize AI benefits, enabling broader societal advancement.

Finally, there's a broader ethical discussion about the role of major tech companies in shaping AI's future direction. With the power to set norms and standards, Google's deployment of AI models like Gemini can influence global digital landscapes significantly. This role comes with the responsibility to consider broader societal impacts, promoting AI development that aligns with ethical standards and societal values. Collaborations with ethical bodies, transparency in AI processes, and accountability measures can ensure that the deployment of AI technologies is both innovative and socially conscious.

The introduction of Google's Gemini 2.5 Pro Preview (I/O edition) has triggered a series of political implications that warrant thoughtful examination. At the forefront is the balance of technological power, as tech giants like Google continue to consolidate significant advancements in AI capabilities. This concentration of influence stirs political debates over the control and governance of artificial intelligence tools, which can have widespread effects on both national and global levels. The concerns over data privacy and algorithmic bias grow as AI models become more integrated into societal applications, creating pressure on policymakers to introduce robust legislative frameworks that safeguard the public interest.

AI‑driven surveillance capabilities pose another layer of political concern. The potential misuse of advanced AI technologies for monitoring and controlling populations is a topic of international import. Governments and international bodies are increasingly called upon to establish clear regulations and guidelines to prevent any overreach or infringement on personal freedoms. Such proactive measures are critical in ensuring that AI's growth does not come at the expense of civil liberties and privacy rights.

Moreover, the role of AI in influencing political narratives and shaping public opinion cannot be overlooked. The advanced video understanding capabilities of Gemini 2.5 Pro highlight potential risks associated with deepfake technology and misinformation campaigns. The political ramifications of using AI to distort information or sway electoral outcomes underscore the need for transparent AI development processes and international agreements on ethical AI deployment, akin to the standards outlined in articles like [Google news](https://techcrunch.com/2025/05/06/google‑debuts‑an‑updated‑gemini‑2‑5‑pro‑ai‑model‑ahead‑of‑i‑o/).

Furthermore, the rise of AI models like Gemini 2.5 Pro accentuates the digital divide between nations with advanced technological infrastructures and those without. This disparity necessitates cooperative international efforts to promote equitable access to AI resources and knowledge sharing. Collaborative initiatives, aimed at addressing these inequalities, can help bridge gaps and foster a more balanced global technological ecosystem, reflecting sentiments shared by many experts in the field. The ongoing advancements and applications of AI require not just technological evolution but also parallel developments in public policy and international relations.

The launch of Google's Gemini 2.5 Pro Preview (I/O edition) represents a significant milestone in the evolution of AI technology, especially in the realm of coding and interactive web application development. This updated model's release reflects Google's commitment to advancing AI capabilities while simultaneously responding to user feedback and demands. With enhancements in coding capabilities such as web app development, code transformation, and code editing, the model is poised to streamline and enhance development processes. These improvements are crucial as they align with the growing trend of AI integration across various sectors, providing developers with more robust and versatile tools to accelerate innovation .

The release ahead of the Google I/O developer conference strategically positions Gemini 2.5 Pro Preview as a harbinger of the advancements expected to be showcased at the conference. By unveiling this model, Google not only builds anticipation for further AI‑related achievements but also underscores the competitive landscape in which tech giants are relentlessly pushing for superior AI solutions. This context is creating a momentum that is likely to catalyze innovation across the industry, as companies seek to outdo each other in terms of capability and performance .

The model's state‑of‑the‑art performance in video understanding tasks, as evidenced by its high score on the VideoMME benchmark, indicates a promising future for AI applications in media processing and analysis. However, with these advancements come discussions around the ethical implications of such technologies. The capacity for video processing could potentially lead to the misuse of AI for creating deepfakes or facilitating misinformation campaigns, thus calling for stringent ethical standards and policies to govern AI deployment .

Economically, the implications of the Gemini 2.5 Pro Preview's enhanced capabilities could result in cost efficiencies and productivity gains for businesses. With the ability to accelerate software development cycles and reduce labor costs, companies could enjoy increased profitability. This potential broadens the model's appeal not just among developers but also in various industries where AI can drive innovation and efficiency. However, it is imperative for Google to address the initial errors and performance setbacks reported by early adopters, as these could dampen the enthusiasm of potential users and influence overall adoption rates .

The public reaction to the Gemini 2.5 Pro Preview has been largely positive, particularly due to its improved coding features which have resonated well within developer communities. Nonetheless, the user‑reported issues such as internal errors and slow performance indicate areas that require immediate attention to ensure the model's reliability and effectiveness. Google's proactive engagement with the developer community will be crucial to resolve these concerns promptly and maintain the momentum of positive reception .

Conclusively, the release of the Gemini 2.5 Pro Preview is a reflection of Google's strategic vision for AI development and integration. It presents opportunities for economic gains and societal advancements but also poses challenges that require thoughtful governance and regulation. By addressing the existing issues and leveraging the strengths of the model, Google can assure widespread adoption while advocating for responsible and ethical AI practices on a global scale .