Laws and administrative rules have been addressing disability and accessibility, through quotas and financial penalties, or students social criteria assessment on French national ranking platforms. Such official measures enable minorities and other discriminated groups to be represented.As half of the humanity does not constitute a minority, no quota policy is supposed to be applied, leaving gender imbalance as a potential issue. Academics have launched various inquiries and studies on this issue (Mothers in Science, UNESCO I’d Blush if I Could, etc.), that is crucial because of the female under-representation in scientific fields in STEM (Science, Technology, Engineering, and Mathematics).“Gender equality paradox”names the observation that the under-representation of women in scientific fields (in particular, those related to STEM) is stronger in the most developed countries. As one says: “We are open. The door is just very heavy”. The door is not closed but women must get a foot in the door to be invited, integrated, accepted, and hired. Women fight to break the glass ceiling in many professions. At the same time, as they can encounter obstacles, the sticky floor restricts them to relative non-strategic positions and prevents their scientific career from really taking off.The tooth paste tube phenomenon is also invoked as a metaphor that illustrates how few female colleagues must assume the committees and representations for the community to comply with F/M quotas when required for hiring or mediating, for instance.The consequence can be an over-solicitation and over-impact on the communities. The facts are horrendous, and actions have hence to be taken to reach a global balance with diversity, inclusiveness and fair cohabitation without discrimination. All the launched initiatives testify this awareness. As a warning underlies the need to support these actions, one must take care of a negative effect on under-represented members by (unfairly) reinforcing their impostor syndrome.Inside the science itself, and not only in terms of population or scientometrics, but a new phenomenon also arises with the evolution of topics, technics, and tools. One striking example is the risk with gender and minority “invisibilization”in AI, as the social representation biases are emphasized (few entries in Wikipedia, Matilda effect, lack of historical figures and illustration, rare “role models”in science,...) with unbalanced learning data. The dual issue is how rule-based systems and by whom, if any bias here also, are encoded to be aware of all the diversity any decision implies.