The Role of Artificial Intelligence in Political Advertising and Crisis Communication: A Case Study of AI-Generated Speech of a Political Leader

Artificial Intelligence has become a basic necessity in 20th century with the emergence of new AI tools and APPs. As artificial intelligence has occupied the functionality of major human based activities, its integration is also seeing a transformative shift in political advertising, especially in the creation of speeches and in communicating party agendas to the general public and plying a vital role in opinion making and agenda setting. AI is already being used all over the globe to tailor the political speeches keeping in view the interests of specific voter segments. Recently the use of AI in political advertising has seen a dramatic shift when Pakistan's imprisoned former Prime Minister Imran Khan delivered an AI generated cloned speech that has stunned the nation. The AI-generated audio, which was played during a virtual rally of his Pakistan Tehreek-e-Insaf (PTI) party, featured a voice replicating Khan's, praising his supporters and urging people to vote for PTI in the upcoming general elections. This innovative use of technology has the potential to impact the political landscape for the upcoming general elections in Pakistan. In regard with the case study of Imran Khan’s AI generated speech, the aim of this research is to explore the effectiveness of his message conveyed through AI and its overall impact on the manipulation of public opinion and election campaign. This paper investigates public reactions and evaluates the effectiveness of delivering personalized messages through AI-driven content. This research also addresses the concerns related to privacy, manipulation, and the need for transparency in the development and deployment of AI-driven political advertising. This research study is conducted through a mixed-method approach, combining qualitative and quantitative research. Research data is obtained from the surveys and interviews of political analysts. Based on the results of research, it will be presenting to which extent AI should be used, what ethical considerations should be met and how a regulatory frame work should be there to keep an eye on the AI generated content.

Artificial Intelligence has become a basic necessity in 20 th century with the emergence of new AI tools and APPs.As artificial intelligence has occupied the functionality of major human based activities, its integration is also seeing a transformative shift in political advertising, especially in the creation of speeches and in communicating party agendas to the general public and plying a vital role in opinion making and agenda setting.AI is already being used all over the globe to tailor the political speeches keeping in view the interests of specific voter segments.Recently the use of AI in political advertising has seen a dramatic shift when Pakistan's imprisoned former Prime Minister Imran Khan delivered an AI generated cloned speech that has stunned the nation.The AI-generated audio, which was played during a virtual rally of his Pakistan Tehreek-e-Insaf (PTI) party, featured a voice replicating Khan's, praising his supporters and urging people to vote for PTI in the upcoming general elections.This innovative use of technology has the potential to impact the political landscape for the general elections in Pakistan and more specifically in time of crisis communication.In regard with the case study of Imran Khan's AI generated speech, the aim of this research is to explore the effectiveness of his message conveyed through AI and its overall impact on the manipulation of public opinion and election campaign.This paper investigates public reactions and evaluates the effectiveness of delivering personalized messages through AI-driven content.This research also addresses the concerns related to privacy, manipulation, and the need for transparency in the development and deployment of AI-driven political advertising.This research study is conducted through a mixed-method approach, combining qualitative and quantitative research.Research data is obtained from the surveys and interviews of political analysts.Based on the results of research, it will be presenting to which extent AI should be used, what ethical considerations should be met and how a regulatory frame work should be there to keep an eye on the AI generated content.

Introduction
In computer science, Artificial Intelligence (AI) is the study of machines that are capable of learning, reasoning, solving problems, seeing, and comprehending language-tasks that normally require human intelligence.Tucker et al. 2018 explored that AI has been used for the targeted advertising and micro-targeting strategies, by analysing the data to identify specific demographics and tailor political messages accordingly.(Holmes et al., 2019) stated that automated technologies, have been deployed to generate political content, like political speeches and advertisements.This research concluded that the use of NLP algorithms was used to create persuasive political messages.Diakopoulos (2019) delves into the ethical implications of personalized and highly targeted political ads.This study raised the concerns related to privacy and the manipulation of public opinion.(Li et al., 2020) mentioned that the rise of deepfake technology has also significantly influenced political advertising, prompting studies that investigate the impact of deepfakes on public perception and the potential for deception in political campaigns.Larsson et al. (2021) explored that the intersection of AI and sentiment analysis and assessed the accuracy and reliability of gauging public sentiment towards political candidates and issues.
Because of developments in generative AI, artificial intelligence has become increasingly important in political advertising.Recent research demonstrates how AI technologies, including generative language models, are applied to provide customised and convincing political messaging on a large scale.These tools improve microtargeting efforts by producing text, photos, and videos that are customised for particular audience segments.Campaigns can benefit from AI-generated content by using less labour and money than traditional advertising approaches, which makes it possible for campaigns with limited resources to compete successfully (Brennan Center for Justice, 2023).AI-generated political speeches are becoming more common in political campaigns.For example, ChatGPT has been used to produce first drafts of speeches and campaign marketing materials.Representative Jake Auchincloss was the first member of Congress who deliver an AI-written speech on the floor of the House of Representatives.The speech was written using ChatGPT and was in support of a bill that would establish a joint AI research center run by the US and Israel.There are several examples of AI-generated speeches being used in politics.In 2016, Valentin Kassarnig at the University of Massachusetts, Amherst, created an artificial intelligence machine that learned how to write political speeches by using a database of almost 4,000 political speech segments from 53 U.S. Congressional floor debates to train a machinelearning algorithm to produce speeches of its own mentioned in MIT technology Review website arXiv, E. T. F. T. (2016, January 19).In Argentina, former President Mauricio Macri gave a speech at an event celebrating the 35th anniversary of Fundación Libertad, which was generated by generative AI tools such as Open AI's ChatGPT.
These examples highlight a worldwide change in political communication, with AIgenerated speeches paving the way for the broader impact of artificial intelligence in political advertising.This trend extends to upcoming elections, such as Pakistan's 2023 General election, where AI, particularly ChatGPT, is playing a pivotal role in reshaping campaign strategies and influencing voter perceptions.However, the use of AI-generated speeches in politics has sparked concerns regarding legality and ethics.During a virtual rally for Pakistan's Tehreek-e-Insaf (PTI) on December 17, 2023, Imran Khan's Digital Media Team utilized an AI-crafted audio clip with an accompanying AI-generated image of Khan addressing supporters.The event, viewed by millions on social media platforms, raised questions about the impact of AI in political advertising.The PTI claimed substantial engagement, with 1.4 million on Facebook, 1.2 million on YouTube, and 1.5 million in a space organized on X.This research aims to assess the effectiveness of AI technology in shaping voter decisions and opinions.
Political campaigns play a vital role in public opinion making, and voting decision.This research paper delves to find the effectiveness, trustworthiness and political implications of using AI in political advertising.Researcher examined the delicate balance between using AI for effective communication, crisis communication and preserving the integrity of democratic debate, keeping in view, the upcoming general Elections in Pakistan and the impact of AI on the political campaigns.The integration of artificial intelligence (AI) in political advertising has the potential to enhance campaign effectiveness, but it also raises concerns about the spread of misinformation and manipulation.This research underscores the pressing need to tackle challenges and risks linked to the growing utilization of AI in political campaigns and the need for a regulatory framework to guide the ethical use of AI in political advertising by drawing insights from global examples of policies and guidelines.The research will contribute to the ongoing debate on the regulation of AI in political advertising and inform policymakers, civil society organizations, and the public on the potential benefits and drawbacks of AI in political campaigns and elections.

AI in Political Advertising
The advanced level of AI-generated content-including speeches and advertisementsallows for real-time and personalised interactions with voters.Political messaging can have a greater impact if content is created using these technologies to speak to the interests and concerns of certain voters (Woolley & Howard, 2018).But there are also moral questions about the application of AI in political advertising.Critical difficulties that must be addressed are data privacy, algorithmic biases, and the possibility for misinformation.To reduce these hazards, ethical standards and transparency in AI operations are essential (Chessen, 2020).
Furthermore, the adoption of AI in political advertising has completely altered how advertising campaigns are run.AI-driven data analytics make it possible to create highly targeted ads that are more effective and engaging because they can be customised to fit into particular demographics.Recent research highlights AI's capacity to evaluate massive datasets in order to forecast voter behaviour and modify messaging rightly, increasing the accuracy of political outreach (Kreiss, 2022).AI tools, for example, have been used to generate campaign speeches as well as laws, demonstrating their versatility and impact.These initiatives have had varying degrees of effectiveness in swaying voters and handling emergencies, though.Because AI tools are effective and scalable, some political campaigns have welcomed them; yet, others are wary of the ethical and reputational concerns involved (Centre for Media Engagement, 2023).
There are a number of moral and practical issues with AI being used in political advertisements.AI-generated content has the potential to spread false information and occasionally be boring or repetitive.Ensuring the integrity and truth of political communications is particularly hampered by the quick creation of content (Goldstein et al, 2023).Further research is needed to determine how AI-generated political advertisements affect voter behaviour, as some studies indicate that they have little effect on shifting voter preferences but a great deal of potential to distort perceptions and strengthen preexisting biases (Brookings Institution, 2023).

AI in Crisis Communication
The transition to holistic crisis communication-which incorporates multiple facets of crisis management into a more cohesive strategy-is covered in the proceedings of the 2024 International Crisis and Risk Communication Conference.The literature study emphasises that in order to improve real-time communication and decision-making, stakeholders must coordinate and leverage cutting-edge technologies This all-encompassing approach seeks to enhance both short-term reactions and long-term recuperation.(International Crisis and Risk Communication Conference, 2024).
There are several difficulties with using AI in crisis communication, though.AIgenerated messages' veracity and authenticity are a source of concern.Because AI has the ability to spread false information during emergencies, its use must be governed by strong legal frameworks and moral principles (Simchon et al, 2024).Johansson and Vigsø (2020) investigate how political controversies might be managed through strategic communication, highlighting the significance of public perception and media dynamics.They list important response techniques-like denying the incident, assigning responsibility, or expressing regret-and examine how these are used in relation to the magnitude and political climate of the crisis.The writers stress the need of timely and consistent communications in managing the narrative, preventing harm to one's reputation, and regaining the public's trust.Their work emphasises how important the media is to crisis communication during political scandals, serving as both a platform and a challenge.
The crucial function of a crisis communications team in successfully handling organisational crises (O'Rourke & Smith, 2023).They emphasise the necessity of prompt, wellcoordinated answers and unambiguous messaging, especially in the rapidly evolving digital environment.The paper emphasises how strengthening crisis communication techniques requires scenario design, ongoing training, and the adoption of new technology.

Methodology
This study employs a mixed-methods approach to thoroughly investigate the intricacies of AI-generated political advertising in realm of crisis communication, specifically examining Imran Khan's AI-generated speech during a virtual PTI party rally.The research incorporates a case study methodology, integrating both qualitative and quantitative components.In the qualitative strand, the data was collected through individual interviews of 2 political communication experts telephonically to capture expert insights on the effectiveness of Imran Khan's AI cloned voice and its impact on the election campaign while party being banned from all the media channels.The interviews were conducted on three basic themes, impact of the AI generated speech on the election campaign, impact on voters and democratic process.On the quantitative side, data was collected through a structured survey to evaluate social media users' perspectives on AI-generated political speech of Imran khan, its impact on freedom of expression, manipulation of AI generated content, impact on voting decisions and AI's role in agenda setting.The questionnaire featured closed-ended questions to facilitate quantitative analysis.The sampling strategy involved a representative snowball sample drawn from a diverse population voters.This approach ensures the generalizability of findings across various demographic and political backgrounds.Administered online through Google Forms, the survey aimed to reach a diverse array of respondents, providing a comprehensive understanding of the multifaceted implications of AI-generated political communication.The results of the study cannot be generalised because of the small population sample and the case study of a specific event.

Impact of Imran Khan's AI generated cloned Speech on the election campaign
This is first time in the history of Pakistan and overall, in Asia that an abducted party leader's voice was cloned and he addressed his followers from jail.Chairman Pakistan Tehreek-e-Insaf is imprisoned for the past six months facing multiple charges.Meanwhile Pakistan's general elections were announced on 8 th February 2024 and the party was banned from conducting any election campaign and rallies physically.Here Artificial Intelligence comes to their rescue and party social media team achieved the milestones by engaging the party voters online through continuous information flow, online jalsas and produced an Imran Khan written speech where he addressed the nation and encouraged them to vote for his party.
The thematic analysis of the interviews suggested that Imran Khan's AI generated speech will surely impact the election campaign whether it's the AI generated or the clips of his old speeches as his followers definitely needs a motivation from their leader.It suggested that social media platforms and AI generated clone voices or animated videos will be a game changer for PTI especially when they are facing restrictions in running the physical election campaign.The party would only have this option left to utilise AI generated material for their campaigning and to engage their supporters.AI has made it impossible to keep PTI out of the electoral process.PTI supporters believe in Imran Khan and they are ready to accept anything coming from him in any version.However, the importance of physical presence will remain there because people can connect to the person talking.Human behaviour and human psychology are not that advanced that it easily could accept the AI generated content.This is attractive but it will take a little time to absorb AI generated clones.

Impact of Imran Khan's AI generated cloned Speech on voters
The analysis suggested that the dynamics of voting decision is entirely different in Pakistan.Voting is a very complex decision and there are multiple factors involved.People have strong party affiliations, peer pressure, social and moral grounds which influence their voting decision.Only the AI generated content would not be enough to change the voting behaviour of people.Voter would only be impressed by the party leader and his narrative; this kind of gimmicks can't fascinate him.Hence Imran Khan's AI generated speech was definitely a ray of hope for people who support him but this will not impact the voting decision of people.

One of the interviewees responded;
"This speech will not impact the voters but definitely this will impact the supporters of Imarn khan because they were eagerly waiting to listen to their leader.His supporters don't have any issue if they are listening to him directly or through AI" Second Respondent; "Only the swing voters would be affected by the AI generated Messages and these are those voters who usually decide on last days, keeping in view the momentum of the election"

Impact of Imran Khan's AI generated content on democratic process
According to the data, AI-generated content can exert dual effects on the democratic process.On the favourable side, it enables politicians to disseminate their messages to a broader audience, aiding in the construction of their narratives.Additionally, it can facilitate crisis communication, particularly when hurdles arise in conveying messages to the public, as evidenced by PTI's utilization of online election campaigns following bans.On the downside, opponents could potentially exploit AI as a propaganda tool to undermine their rivals' narratives and reputations, leading to a state of uncertainty.However, AI has the capacity to enhance message productivity by tailoring content to the demographics of the audience, thereby facilitating effective agenda-setting by politicians.This has the potential to positively influence the democratic system, as voters can make informed decisions based on the content presented.Nevertheless, the proliferation of fake news poses a significant challenge due to the ambiguity surrounding intellectual property rights.With the ease of manipulating audio and video content, AI presents a dangerous avenue, exacerbated by the lack of public awareness and adequate legal frameworks.Fake news could be wielded to disrupt and manipulate election campaigns, posing a threat to the democratic process.Therefore, the legal mechanisms designed to counter propaganda should extend to encompass AI-generated political content aimed at influencing election campaigns.

Survey Findings and analysis
The survey was responded by 200 people with different party affiliations.Among them 35.6% were PTI supporters.The age group who answered were from 20 to 46 years of age, 17.8% were of 22 years of age.75.5 % listened to the Imran Khan's AI generated speech.If we analyse this data, it suggests that people listened to his speech were not only the PTI supporters but were also from other parties.95.6% listened the speech directly on social media platforms.Approximately 94.7% considered that AI could effectively be used as tool to interact with the target group, if a party is facing any sanctions in public speaking.When asked about the speech effectiveness only 46.7% found it convincing which is a little higher than the percentage of PTI supporters who took part in the survey.So assumed that the most of the people who found the speech persuasive and effective were PTI supporters.
Graph No 1: Respondents from different party affiliations 48.9% respondents answered that they would vote for Imran Khan after watching his AI generated speech.This data again shows that the figure is closer to the supporters of PTI which was 35.6%.Only 13.3% were the new voters who decided to vote for him after listening to his AI generated speech which is not a large number.
According to data people were confused about the originality of the message.Only 28.9% answered that the speech was original words of Imran Khan whether, 51.1% were not sure about the authenticity of the message.The percentage who agreed is even less than the percentage of supporters of PTI.Therefore, it is evident that people have doubts about the authenticity of the content developed through artificial intelligence.
Graph No 2: Words used in AI Speech 37.8% respondents replied that they would trust the cloned image or voice of their candidate if they have a stronger affiliation with the party leader.This data shows that the AI generated clones and content would only be recognized and accepted by the supporters of a specific party and there is a very little margin left for the new people to get impressed by AI generated material.
40% of the respondents replied that they would not vote for Imran Khan after listening to his AI generated speech, even if they have no affiliation with any other party or candidate.53.3% responded that candidates must avoid such AI generated clones in future.33.3% found AI-generated personalized political messages effective in terms of persuading someone to vote for a particular candidate.71.1% are of the view that physically delivered speeches are more effective and convincing that the one generated through AI.
Graph No 3: Effectiveness of AI Speech 44.4% were of the view that they would vote for a candidate only if they had a political connection with that party not because of the impact of AI generated messages.

Graph No 4: Effect of AI Speech regarding decision about Vote
51.1% of the sample population expressed its apprehension that false cloned voices or characters of political candidates could be used as propaganda tools to manipulate the general elections in Pakistan.
Graph No 5: AI Speech used as used as propaganda tools

Conclusion
AI-generated speech is changing how we communicate and present ideas.This technology involves feeding information like the speech topic, tone, and audience to an AI algorithm, which then generates the speech.This can revolutionize communication by providing high-quality speeches quickly, without the need for extensive writing and rehearsal.In the realm of politics, where public speaking and advertising play a big role, AI is transforming how campaigns are executed, assessed, and targeted.It can be really beneficial for the political leaders and parties to engage the targeted voters in any crisis situation.AI's ability to target specific voter groups and improve ad placement and content has the potential to reshape political advertising.It has also been observed that AI-generated voices mostly lose the human touch.AI-generated speech can lack the personal and emotional connection that is common in speech created and performed by humans.This can make it difficult for the audience to interact with the speaker and understand the message being conveyed.As the Imran khan's AI generated voice was not exactly the same and it pronounced most of the words wrong.The study concludes that there is an insignificant impact of AI in political advertising on public perception and trust.The reason behind is the authenticity and transparency in personalized political messages.Research shows that people are concerned about the manipulation and fake messages which ultimately restrict them to be influenced by AI generated content.Since, no rules and regulations exist regarding digital media policies in Pakistan; massive use of AI in political advertising can potentially damage the electoral processes.The UK's Information Commissioner's Office (ICO) has developed a general accountability framework that provides a baseline for demonstrating accountability under the UK GDPR, which can be applied to AI accountability.This framework includes aspects such as fairness in the AI lifecycle and the need to consider competing interests when assessing AIrelated risks.Additionally, the European Parliament emphasizes the importance of transparency, accountability, fairness, and regulation in the introduction of AI to build trust and respect human and civil rights.Furthermore, the concept of "Fairness, Accountability, Transparency, and Ethics" (FATE) has gained interest in AI education, highlighting the significance of these principles in AI development and implementation.

Implications of the study
In light of worldwide apprehensions and standards regarding AI frameworks, this research will be instrumental as it offers a framework for striking a balance between ethical considerations and technological improvements.It will serve as a guide for legislators as they create regulations that address the dangers of false information and uphold public confidence.The study's focus on authenticity and openness will guide best practices for AI-generated content, guaranteeing that political messages are interesting and believable.Furthermore, the study will contribute to the continuing debates concerning AI's place in democratic processes by helping to develop tactics that protect election integrity and promote educated public discourse.This research will contribute to the development of a more responsible and moral application of AI in political communication by emphasising the need for cooperation between technology developers, legislators, and civil society.